Metraipolis › archive › dc_39b95fcb77d2

Stable Geometry with Divergent Task Evidence for Efficient Long-Horizon Agent Compression — arXiv:2609.27332

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.27332
License (as archived): CC0-1.0
Archived: 2026-10-08T09:11:22.251Z
SHA-256 of the archived text: c18217d85d47bfabd27389b87b7643c0696f2787aeaea3706f5d62e209368564
Stable URL of this page: https://metraipolis.com/documento/dc_39b95fcb77d2

How to cite

Stable Geometry with Divergent Task Evidence for Efficient Long-Horizon Agent Compression — arXiv:2609.27332 — arXiv (en), https://arxiv.org/abs/2609.27332 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_39b95fcb77d2 (Metraipolis archive dc_39b95fcb77d2, sha256 c18217d85d47bfabd27389b87b7643c0696f2787aeaea3706f5d62e209368564)

BibTeX

@misc{dc39b95fcb77d2,
  title = {Stable Geometry with Divergent Task Evidence for Efficient Long-Horizon Agent Compression — arXiv:2609.27332},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.27332},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_39b95fcb77d2; sha256 c18217d85d47bfabd27389b87b7643c0696f2787aeaea3706f5d62e209368564},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27332. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long horizon agents accumulate growing interaction histories that increase context and inference costs. We find that geometric redundancy alone is an insufficient criterion for safe compression. Although agent histories exhibit strong low dimensional structure, similar global geometry can preserve very different amounts of task evidence. At identical retained block counts, evidence aware selection raises next action Top 3 retention from 0.31 to 0.69, while centroid similarity remains 0.98. Controlled replacement further shows that action related information can be substantially altered while global geometric measures remain nearly unchanged. Motivated by this gap between geometry and evidence, we introduce Geometry Guided Evidence Preserving Memory (GEM), a training free compressor that protects task and execution evidence before using geometric residuals to complete coverage. GEM reduces mean combined token usage from 2.69M to 2.11M per task, a 21.4% reduction, while maintaining comparable task reward. Our results show that efficient agent history compression should optimize for preserved task evidence rather than geometric coverage alone.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.27332 — 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 2026-10-08T08:13:26.502Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).