Metraipolis › archive › dc_b97b76a2a3fd

AlterNet: Adaptive-Depth Mixture-of-Experts with Learned Halting for Elastic Inference — arXiv:2611.08741

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.08741
License (as archived)
CC0-1.0
Archived
2026-11-09T22:33:42.060Z
SHA-256 of the archived text
2afe8016cab0239c64780c84dc005f485f61e32c887868904c1b9ec34a7350a4
Stable URL of this page
https://metraipolis.com/documento/dc_b97b76a2a3fd

How to cite

AlterNet: Adaptive-Depth Mixture-of-Experts with Learned Halting for Elastic Inference — arXiv:2611.08741 — arXiv (en), https://arxiv.org/abs/2611.08741 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b97b76a2a3fd (Metraipolis archive dc_b97b76a2a3fd, sha256 2afe8016cab0239c64780c84dc005f485f61e32c887868904c1b9ec34a7350a4)
BibTeX
@misc{dcb97b76a2a3fd,
  title = {AlterNet: Adaptive-Depth Mixture-of-Experts with Learned Halting for Elastic Inference — arXiv:2611.08741},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.08741},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b97b76a2a3fd; sha256 2afe8016cab0239c64780c84dc005f485f61e32c887868904c1b9ec34a7350a4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.08741. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Most LLM architectures have fixed compute demands determined at training time. We propose a novel architecture that combines the Mixture-of-Experts architecture with shallow experts and a novel dynamic-depth system to make the extent of a forward pass dynamically adjustable. Our approach lets the model automatically determine the compute needed for a given task, paired with a system to scale it down after training without retraining when the execution environment demands faster or cheaper inference. To achieve this, we introduce an iterative computation mechanism, sequence-level depth-based routing steered by a halting mechanism within the experts, a depth encoding that lets the model track its progress in the chain, and a regularization mechanism that incentivizes the model to use allocated compute efficiently. We present initial benchmarks and discuss future research directions.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2611.08741 — 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 2026-11-09T22:30:47.254Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).