Metraipolis › archive › dc_56b95d43e50d

Efficient Reinforcement Learning for Long-Horizon Tool-Use Agentic Tasks — arXiv:2608.10357

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.10357
License (as archived)
CC0-1.0
Archived
2026-10-19T06:09:04.737Z
SHA-256 of the archived text
635cb64fa0993d34f601a213256a316230d1d91a5b5c89651bfb758899101515
Stable URL of this page
https://metraipolis.com/documento/dc_56b95d43e50d

How to cite

Efficient Reinforcement Learning for Long-Horizon Tool-Use Agentic Tasks — arXiv:2608.10357 — arXiv (en), https://arxiv.org/abs/2608.10357 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_56b95d43e50d (Metraipolis archive dc_56b95d43e50d, sha256 635cb64fa0993d34f601a213256a316230d1d91a5b5c89651bfb758899101515)
BibTeX
@misc{dc56b95d43e50d,
  title = {Efficient Reinforcement Learning for Long-Horizon Tool-Use Agentic Tasks — arXiv:2608.10357},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.10357},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_56b95d43e50d; sha256 635cb64fa0993d34f601a213256a316230d1d91a5b5c89651bfb758899101515},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.10357. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long-horizon tool-using agents must reason over user goals, domain policies, tool calls, simulator state, and delayed verifiable rewards. Reinforcement learning (RL) is a natural fit for this setting, but multi-turn on-policy rollouts create long contexts, while model-specific attention layers may require custom masks and learned sink normalization. We present SINKFLEX-RL, a modular training system for RL in dual-control tool-use environments. The system combines a Gymnasium-compatible environment wrapper, a VERL-style rollout dataflow, group-relative policy optimization without a separate value model, and a sink-aware FlexAttention path designed to preserve model-specific sink scaling under causal and sliding-window masks. In a preliminary Tau2Bench retail run, validation reward (mean@1) rises from 0.25 early in training to $0.44$ later in the observed training window, while training-score and trajectory-reward proxies also trend upward. In a fixed-configuration memory benchmark, the optimized attention path reduces peak VRAM from 28.06GB to 22.52GB at 4096 tokens, a $19.7\%$ reduction, and runs the measured 8192-token configuration using $25.53$~GB where the eager baseline runs out of memory. These results illustrate the value of integrating environment interfaces, RL dataflow, and attention-kernel design for memory-feasible long-horizon agent training.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-12 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.10357 — 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 2026-10-19T05:57:28.354Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-12?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).