Metraipolis › archive › dc_78ebe8691aef
What Should an Agent Remember? Disentangling Retention from Retrieval in Bounded-Memory Evaluation — arXiv:2610.00366
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.00366
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T12:01:57.102Z
- SHA-256 of the archived text
33efc0f9e1576f6b7a58ad41941dc8f9433b760104d3b71126c910a3ddbd2580- Stable URL of this page
https://metraipolis.com/documento/dc_78ebe8691aef
How to cite
What Should an Agent Remember? Disentangling Retention from Retrieval in Bounded-Memory Evaluation — arXiv:2610.00366 — arXiv (en), https://arxiv.org/abs/2610.00366 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_78ebe8691aef (Metraipolis archive dc_78ebe8691aef, sha256 33efc0f9e1576f6b7a58ad41941dc8f9433b760104d3b71126c910a3ddbd2580)
BibTeX
@misc{dc78ebe8691aef,
title = {What Should an Agent Remember? Disentangling Retention from Retrieval in Bounded-Memory Evaluation — arXiv:2610.00366},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.00366},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_78ebe8691aef; sha256 33efc0f9e1576f6b7a58ad41941dc8f9433b760104d3b71126c910a3ddbd2580},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.00366. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
A persistent agent must decide both what to retain as information arrives and what to surface once a query appears, yet memory evaluations can confound these decisions by comparing methods that differ in both retention and selection. We build a streaming-recall benchmark crossing retention and selection rules and evaluate every condition on the same 300 seeded episodes. Holding access fixed, query-aware selection improves required-fact recall by 15.5 percentage points (95% CI: 12.8 to 18.2), whereas a mixed comparison that also changes history access reports a 68.7-point advantage, of which 53.2 points are attributable to access. Under bounded retention, query-aware, dense, and oracle selection reach the retention ceiling, and all 319 observed failures in the bounded recency condition are caused by eviction rather than ranking errors. Recall falls to 0% as targets recede sufficiently far into the past. Repeating the evaluation on SQuAD preserves the retention ceiling while showing that dense retrieval can outperform lexical retrieval on natural text. These results show that bounded-memory evaluations should hold access fixed and report retention and selection separately. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-02 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2610.00366 — 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 2026-10-06T10:21:27.861Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).