Metraipolis › archive › dc_95e4e51d3468
GTRL: Grounding Divide-and-Conquer Value Learning with Temporal Differences — arXiv:2609.33259
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.33259
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T22:37:19.909Z
- SHA-256 of the archived text
6b12c65523ca790f6056ff16b6adbc6560c260f0ce167c8558cb133d3143c7a7- Stable URL of this page
https://metraipolis.com/documento/dc_95e4e51d3468
How to cite
GTRL: Grounding Divide-and-Conquer Value Learning with Temporal Differences — arXiv:2609.33259 — arXiv (en), https://arxiv.org/abs/2609.33259 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_95e4e51d3468 (Metraipolis archive dc_95e4e51d3468, sha256 6b12c65523ca790f6056ff16b6adbc6560c260f0ce167c8558cb133d3143c7a7)
BibTeX
@misc{dc95e4e51d3468,
title = {GTRL: Grounding Divide-and-Conquer Value Learning with Temporal Differences — arXiv:2609.33259},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.33259},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_95e4e51d3468; sha256 6b12c65523ca790f6056ff16b6adbc6560c260f0ce167c8558cb133d3143c7a7},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33259. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
In offline goal-conditioned reinforcement learning (GCRL), divide-and-conquer scales to long horizons by joining two shorter segments at a subgoal. However, under stochastic dynamics, the base case of this rule values the luckiest trajectories through the data. The subgoal must also lie on a shared trajectory, so a state-goal pair that no trajectory connects gets no value update at all. To address both, we present Grounded Transitive RL (GTRL), an offline GCRL value learning algorithm that grounds the divide-and-conquer update with a one-step TD target. Over a single step, TD is correct, as its target averages over the successors and needs no subgoal. GTRL adds this target to the composition rather than replacing it, so every pair receives an update, and the composition still carries the long horizon. GTRL also corrects the bias from hindsight relabeling by reweighting each goal against how reachable it was from other successors. We evaluate our algorithm on nineteen OGBench tasks spanning stochastic, deterministic, and stitching environments, where it achieves the highest average success rate. Code will be released soon. 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/2609.33259 — 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 2026-10-07T22:14:27.911Z 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).