Metraipolis › archive › dc_be9d4168cf8a

Scalable In-Context Reinforcement Learning with Recurrent Algorithm Distillation — arXiv:2609.35333

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.35333
License (as archived)
CC0-1.0
Archived
2026-10-07T22:39:02.929Z
SHA-256 of the archived text
7d18361875f036cb782da0489fa61e1825215e31ce2f0144070500887584f38c
Stable URL of this page
https://metraipolis.com/documento/dc_be9d4168cf8a

How to cite

Scalable In-Context Reinforcement Learning with Recurrent Algorithm Distillation — arXiv:2609.35333 — arXiv (en), https://arxiv.org/abs/2609.35333 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_be9d4168cf8a (Metraipolis archive dc_be9d4168cf8a, sha256 7d18361875f036cb782da0489fa61e1825215e31ce2f0144070500887584f38c)
BibTeX
@misc{dcbe9d4168cf8a,
  title = {Scalable In-Context Reinforcement Learning with Recurrent Algorithm Distillation — arXiv:2609.35333},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.35333},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_be9d4168cf8a; sha256 7d18361875f036cb782da0489fa61e1825215e31ce2f0144070500887584f38c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.35333. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Algorithm Distillation (AD) has demonstrated the remarkable ability of Transformers to perform in-context reinforcement learning without explicit weight updates. However, capturing long-term learning progress necessitates expansive context windows, which incur prohibitive memory costs and limit scalability in complex, long-horizon tasks. To address this bottleneck, we propose Recurrent Algorithm Distillation (RAD). RAD employs a dual-component architecture: a Compression Transformer that distills extended interaction histories into compact latent tokens, and an AD Transformer that auto-regressively generates actions using a hybrid context of these compressed memories and recent transitions. By maintaining a fixed-size latent buffer, RAD decouples the effective history length from computational complexity, functionally providing the model with a long-horizon memory. Empirical evaluations across diverse environments demonstrate that RAD matches the asymptotic performance of standard AD with significantly reduced context window sizes, offering a scalable solution for efficient in-context decision-making.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.35333 — 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 2026-10-07T22:14:28.417Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).