Metraipolis › archive › dc_c22ecea994f0
TraceRelay: Attention-Aligned Recurrence over Rolling Traces — arXiv:2610.11743
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.11743
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-09T06:23:28.886Z
- SHA-256 of the archived text
baadc693071a21ac6fada21980bc3be7cde40311cf13c714d9109e53477988fc- Stable URL of this page
https://metraipolis.com/documento/dc_c22ecea994f0
How to cite
TraceRelay: Attention-Aligned Recurrence over Rolling Traces — arXiv:2610.11743 — arXiv (en), https://arxiv.org/abs/2610.11743 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c22ecea994f0 (Metraipolis archive dc_c22ecea994f0, sha256 baadc693071a21ac6fada21980bc3be7cde40311cf13c714d9109e53477988fc)
BibTeX
@misc{dcc22ecea994f0,
title = {TraceRelay: Attention-Aligned Recurrence over Rolling Traces — arXiv:2610.11743},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.11743},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c22ecea994f0; sha256 baadc693071a21ac6fada21980bc3be7cde40311cf13c714d9109e53477988fc},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.11743. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We present TraceRelay, an attention-aligned recurrent architecture that distributes persistent representations over a rolling sequence of low-dimensional traces. Local right looking attention forms increments from lower-layer representations; delivery is delayed until all attended inputs are in the causal past. A fixed additive phase recurrence accumulates the delayed increments, and left-looking attention reads the resulting residual augmented stream. A stride-wise prefix sum supports parallel prefill and bounded-buffer continuation. We study 36 small-model runs on Equal Repeats, bounded Dyck closing-type prediction, and causal Most-Freq generation, using three seeds per setting. At trained length 256, Equal Repeats models with recurrent phase inheritance reach 98.81-99.69% accuracy versus 50.73-51.63% for separately trained variants without inheritance, despite the latter receiving more updates. Accuracy drops sharply at lengths beyond the training range. At the longest evaluated lengths, models with more dimensions in the middle layer's recurrent traces perform better on Dyck (76.34% versus 55.49% close accuracy at length 4096), whereas models with fewer trace dimensions perform better on five-symbol Most-Freq (70.74% versus 55.60% exact generation at length 1024). These contrasting cases motivate further study of how the size of recurrent representations should be chosen for different tasks, without establishing a general rule across tasks or model configurations. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.11743 — 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 2026-10-09T05:50:31.602Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).