Metraipolis › archive › dc_aedcc4376e27

When to Forget: Risk-Controlled Context Adaptation for Long-Horizon Agents — arXiv:2610.22897

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.22897
License (as archived): CC0-1.0
Archived: 2026-10-19T08:42:19.207Z
SHA-256 of the archived text: b2bb2720363f8eb193bb5de14ce0e03f0ec58ce0100d8862f64ed851bd496572
Stable URL of this page: https://metraipolis.com/documento/dc_aedcc4376e27

How to cite

When to Forget: Risk-Controlled Context Adaptation for Long-Horizon Agents — arXiv:2610.22897 — arXiv (en), https://arxiv.org/abs/2610.22897 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_aedcc4376e27 (Metraipolis archive dc_aedcc4376e27, sha256 b2bb2720363f8eb193bb5de14ce0e03f0ec58ce0100d8862f64ed851bd496572)

BibTeX

@misc{dcaedcc4376e27,
  title = {When to Forget: Risk-Controlled Context Adaptation for Long-Horizon Agents — arXiv:2610.22897},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.22897},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_aedcc4376e27; sha256 b2bb2720363f8eb193bb5de14ce0e03f0ec58ce0100d8862f64ed851bd496572},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.22897. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long-horizon agents learn from failure, but the value of a failure record can change as execution progresses. We study when retaining, compressing, or removing the same history block improves subsequent execution relative to full-history KEEP. We introduce RECAP, a framework that combines matched context interventions, execution-relational prediction, and independent policy certification. Matched environment states and rollout seeds measure success and continuation prompt-cost differences; prefix-only relations between a failure block and the current execution propose promising edits. Independent task bundles then select a policy subject to harmful-intervention and non-positive-benefit risk constraints under a fixed measurement protocol. A five-seed longitudinal case holds both the failure block and its summary fixed: compression fails at an early state and becomes a cost-saving intervention at a later state. Across five backbones, RECAP produces positive paired-success saving estimates with backbone-specific reliability-efficiency tradeoffs. On Hunyuan-A13B, it edits 43.24% of retained checkpoints while matching KEEP's observed 91.89% continuation success and reducing unconditional mean continuation prompt cost by 4,619.8 tokens (2.13%). The framework provides an execution-level reference for deciding when to forget: retain history by its value to what comes next.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.22897 — 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 2026-10-19T08:39:10.044Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).