Metraipolis › archive › dc_d87e1fdc7ef1
Attending to Memory Writes Reduces Seed Variability in Recurrent Length Generalization — arXiv:2611.04902
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.04902
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-04T18:54:15.979Z
- SHA-256 of the archived text
52b57bd97775f127f23f3492250d14182f5bba2520be316de80736e7698ec211- Stable URL of this page
https://metraipolis.com/documento/dc_d87e1fdc7ef1
How to cite
Attending to Memory Writes Reduces Seed Variability in Recurrent Length Generalization — arXiv:2611.04902 — arXiv (en), https://arxiv.org/abs/2611.04902 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d87e1fdc7ef1 (Metraipolis archive dc_d87e1fdc7ef1, sha256 52b57bd97775f127f23f3492250d14182f5bba2520be316de80736e7698ec211)
BibTeX
@misc{dcd87e1fdc7ef1,
title = {Attending to Memory Writes Reduces Seed Variability in Recurrent Length Generalization — arXiv:2611.04902},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.04902},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d87e1fdc7ef1; sha256 52b57bd97775f127f23f3492250d14182f5bba2520be316de80736e7698ec211},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04902. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Recurrent memory requires learning useful updates, but attention added to assist learning can also compromise length generalization. We study bounded delta attention: causal attention over uncorrected update proposals whose bounded correction enters the recurrent update. Its implementation in TraceRelay combines additive phase recurrence with attention computed before a parallel scan. On prefix-supervised modulo-5, trained at lengths 1--40, three findings characterize its behavior at length 80. First, upper delta attention attains $70.81\pm8.26$% accuracy across seven seeds, versus $69.21\pm22.42$% for the local-attention recurrent baseline (mean $\pm$ sample SD): a 63.2% reduction in SD with a similar mean. Upper state attention also reduces variation, reaching $74.75\pm6.24$%, so this benefit is not unique to update-level attention. Disabling the upper delta correction retains $70.62\pm8.60$%. Second, blocking gradients at the delta-attention inputs changes accuracy to $65.25\pm12.40$% in the matched seven-seed comparison, increasing SD by 50.2%. It also delays a 99% validation threshold on average among the same six seeds that reach it; the 95% threshold is reached earlier on average with input-gradient blocking across all seven seeds. Third, lower delta attention exhibits substantially less extrapolation degradation than lower state attention: 62.60% versus 19.40%, against a 71.06% baseline. Removing its correction reduces accuracy to 22.56%; its advantage over lower state attention also persists under input-gradient blocking. These observations motivate bounded update-level attention as a recurrent learning scaffold, combining reduced seed variation with less severe degradation in a placement where state attention fails. 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/2611.04902 — 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 2026-11-04T18:35:37.465Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).