Metraipolis › archive › dc_b0ccb206fc81

LoopMTP: A looped transformer guided by latent multi-token prediction — arXiv:2608.03624

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.03624
License (as archived)
CC0-1.0
Archived
2026-10-31T12:31:09.266Z
SHA-256 of the archived text
51dc3712829030ff2a7be43ce60453edd9bd6c09e921eb467ed6f533314c3a8b
Stable URL of this page
https://metraipolis.com/documento/dc_b0ccb206fc81

How to cite

LoopMTP: A looped transformer guided by latent multi-token prediction — arXiv:2608.03624 — arXiv (en), https://arxiv.org/abs/2608.03624 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b0ccb206fc81 (Metraipolis archive dc_b0ccb206fc81, sha256 51dc3712829030ff2a7be43ce60453edd9bd6c09e921eb467ed6f533314c3a8b)
BibTeX
@misc{dcb0ccb206fc81,
  title = {LoopMTP: A looped transformer guided by latent multi-token prediction — arXiv:2608.03624},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.03624},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b0ccb206fc81; sha256 51dc3712829030ff2a7be43ce60453edd9bd6c09e921eb467ed6f533314c3a8b},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03624. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Looped transformers have emerged as a parameter-efficient alternative to scaling depth for strong reasoning. By reusing one stack of layers across $T$ iterations, they attain the effective depth and reasoning capabilities of larger models at a fixed parameter count. Yet existing approaches suffer from latent overthinking and undifferentiated computation, largely because intermediate representations receive no guidance across loops. Multi-token prediction (MTP) supplies exactly the dense, forward-looking supervision the loop is missing. We propose \textsc{LoopMTP}, which links the two through a structural correspondence in latent space: a model that loops $T$ times can anticipate $T$ future tokens. \textsc{LoopMTP} realizes this by softly aligning the hidden state of loop $t$ with the embedding of the token $t$ steps ahead, while a lightweight gate preserves useful information across iterations. \textsc{LoopMTP} improves average accuracy by up to 8.1\% (relative) over the non-looped baseline, with training remaining stable for up to 15 loops.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.03624 — 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 2026-10-31T12:17:37.775Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).