Metraipolis › archive › dc_731bcfef1ae3
Three Steps at a Time: Learning Representations from Action Sequences in Contrastive RL — arXiv:2608.30640
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.30640
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T08:39:18.456Z
- SHA-256 of the archived text
4b0e96f06e93b3097f6004142d3b08a89dc50cd9a5af6191c99f518f9e39e251- Stable URL of this page
https://metraipolis.com/documento/dc_731bcfef1ae3
How to cite
Three Steps at a Time: Learning Representations from Action Sequences in Contrastive RL — arXiv:2608.30640 — arXiv (en), https://arxiv.org/abs/2608.30640 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_731bcfef1ae3 (Metraipolis archive dc_731bcfef1ae3, sha256 4b0e96f06e93b3097f6004142d3b08a89dc50cd9a5af6191c99f518f9e39e251)
BibTeX
@misc{dc731bcfef1ae3,
title = {Three Steps at a Time: Learning Representations from Action Sequences in Contrastive RL — arXiv:2608.30640},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.30640},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_731bcfef1ae3; sha256 4b0e96f06e93b3097f6004142d3b08a89dc50cd9a5af6191c99f518f9e39e251},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30640. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
While self-supervised approaches to reinforcement learning have achieved strong results by learning representations of states and actions, a key open question is the time scale over which actions should be modeled. Departing from the standard formulation relying on single-step actions, we extend contrastive reinforcement learning (CRL), a prototypical self-supervised method, to operate over action chunks, and find that this results in large, pervasive gains across established offline and online benchmarks: +31.7% and +93.1% across 18 and 11 environments respectively. While action-chunking-driven gains are generally explained through the ability to model non-Markovian, temporally extended policies, and to propagate unbiased multi-step returns, interestingly, we find that these arguments only partially apply to CRL. Our empirical studies suggest that, in the context of CRL, an action chunk carries more information about the goal than a single action, measurably improving the critic's representations, and rendering the algorithm significantly more effective. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2608.30640 — 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 2026-10-17T08:13:26.832Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).