Metraipolis › archive › dc_94ea37b18ee2
When Should an Agent Refresh? Adaptive Index Reuse for Efficient Long-Horizon Reasoning — arXiv:2611.01429
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.01429
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T18:24:36.594Z
- SHA-256 of the archived text
f8d492f06b376e3eebfe0997c8d707fd369c75d411aa4571092d1be8df866765- Stable URL of this page
https://metraipolis.com/documento/dc_94ea37b18ee2
How to cite
When Should an Agent Refresh? Adaptive Index Reuse for Efficient Long-Horizon Reasoning — arXiv:2611.01429 — arXiv (en), https://arxiv.org/abs/2611.01429 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_94ea37b18ee2 (Metraipolis archive dc_94ea37b18ee2, sha256 f8d492f06b376e3eebfe0997c8d707fd369c75d411aa4571092d1be8df866765)
BibTeX
@misc{dc94ea37b18ee2,
title = {When Should an Agent Refresh? Adaptive Index Reuse for Efficient Long-Horizon Reasoning — arXiv:2611.01429},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.01429},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_94ea37b18ee2; sha256 f8d492f06b376e3eebfe0997c8d707fd369c75d411aa4571092d1be8df866765},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01429. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Language agents read code, tool output, and dialogue in contexts that grow at every step. Dynamic sparse attention lets a small indexer choose the earlier tokens each query attends to, but the indexer still scores the whole prefix at every layer. Cross-layer reuse methods such as IndexCache rerun the indexer at only some layers, with one fixed pattern for every input. In an agent, the relevant evidence can change abruptly when a traceback or test result arrives, so fresh selections matter more at some steps than at others. We propose Adaptive Index Reuse (AIR), which keeps IndexCache's execution and learns, for each segment of an agent episode, how many layers refresh their selection. A small controller reads signals from completed computation and the latest tool event, chooses a pattern from a nested library, and learns from terminal task reward under an expected compute budget. On Qwen3-4B-Instruct-2507 with a 128K context, AIR improves success over the strongest static schedule at matched cost on SWE-bench Verified (27.0% to 32.4%), Terminal-Bench 2.0, and three $\tau^2$-bench domains, and it lowers time per successful task on all five. Across 50 backbone, context, and benchmark settings, it matches or exceeds that schedule in 44. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2611.01429 — 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 2026-11-03T18:10:08.035Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).