Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Offline Reinforcement Learning with Universal Horizon Models

About

Model-based reinforcement learning (RL) offers a compelling approach to offline RL by enabling value learning on imagined on-policy trajectories. However, it often suffers from compounding errors due to repeated model inference on self-generated states. While geometric horizon models (GHM) alleviate this issue through direct prediction over a discounted infinite-horizon future, they remain challenged in accurately modeling distant future states. To this end, we introduce universal horizon models (UHM), a generalization of GHM that directly predicts future states under arbitrary horizons. Leveraging this flexibility, we propose a scalable value learning method that employs a winsorized horizon distribution to stabilize training by capping excessively large horizons. Experimental results on 100 challenging OGBench tasks demonstrate that the proposed method outperforms competitive baselines, particularly on tasks with highly suboptimal datasets and those requiring long-horizon reasoning. Project page: https://rllab-snu.github.io/projects/UHM/

Hojun Chung, Junseo Lee, Songhwai Oh• 2026

Related benchmarks

TaskDatasetResultRank
Offline Reinforcement LearningOGBench
AntMaze Giant Navigate36
78
Offline Reinforcement LearningOGBench long-horizon reasoning tasks
Cube Triple Play Success Rate56
5
Offline Reinforcement LearningOGBench Noisy Tasks
AntMaze Medium Explore Success Rate89
5
Showing 3 of 3 rows

Other info

Follow for update