Metraipolis › archive › dc_40739563efb6

Rethinking Routing Exploration in Modern Mixture-of-Experts — arXiv:2610.22353

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.22353
License (as archived)
CC0-1.0
Archived
2026-10-16T09:50:58.989Z
SHA-256 of the archived text
3cb0667ff7b626c51bde25066afea89d507c8ce004eff0e91ee055594de569f4
Stable URL of this page
https://metraipolis.com/documento/dc_40739563efb6

How to cite

Rethinking Routing Exploration in Modern Mixture-of-Experts — arXiv:2610.22353 — arXiv (en), https://arxiv.org/abs/2610.22353 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_40739563efb6 (Metraipolis archive dc_40739563efb6, sha256 3cb0667ff7b626c51bde25066afea89d507c8ce004eff0e91ee055594de569f4)
BibTeX
@misc{dc40739563efb6,
  title = {Rethinking Routing Exploration in Modern Mixture-of-Experts — arXiv:2610.22353},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.22353},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_40739563efb6; sha256 3cb0667ff7b626c51bde25066afea89d507c8ce004eff0e91ee055594de569f4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.22353. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Mixture-of-experts (MoE) models have become an important approach for scaling model parameters. Modern MoE architectures increasingly combine two design choices: fine-grained expert segmentation and shared experts. The former increases expressiveness by enabling more expert combinations at a fixed activation ratio, and the latter introduces experts that are active for every token. We identify an exploration dilemma between these designs: shared experts can make routing less exploratory by reducing the variability of token representations. This conflicts with the greater need for exploration induced by fine-grained expert segmentation due to the combinatorially enlarged routing space. Consequently, the additional expressiveness offered by fine-grained segmentation may not be fully realized in practice. MirrorMoE is proposed to reconcile this exploration dilemma by shrinking the exploration space through token-adaptive suppression of unnecessary experts. MirrorMoE consistently improves performance across model scales.

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/2610.22353 — 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 2026-10-16T08:44:18.066Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).