The paper introduces RADT, a decision transformer for offline, reward‑free, goal‑conditioned reinforcement learning that can handle arbitrary avoid‑regions specified at evaluation time. By encoding goals and avoid‑regions as prompt tokens and using an avoid‑region hindsight relabeling strategy, RADT learns reach‑avoid behavior from suboptimal random trajectories without reward or cost supervision. Across 17 tasks and environments, RADT outperforms existing offline goal‑conditioned RL models, achieving significant zero‑shot generalization to novel avoid‑region sizes and counts, and is also applied to a biological cell reprogramming task.
arXiv:2507. 15356v2 Announce Type: replace Abstract: Offline reinforcement learning (RL) learns policies from fixed datasets, thereby avoiding costly or unsafe environment interactions.
arXiv:2602. 05459v2 Announce Type: replace Abstract: Offline goal-conditioned reinforcement learning (GCRL) is typically benchmarked by the best tuned success rate of each method.
arXiv:2608.30406v1 Announce Type: new Abstract: Goal-conditioned reinforcement learning struggles with long horizons when rewards are sparse. While a planner can provide subgoals to guide a low-level...
The paper introduces Cross-Behavior Policy Optimization (CBPO), a method that unifies offline demonstrations and policy-generated rollouts into a single regression loss using a fixed-reference target. CBPO eliminates the need for separate supervised fine‑tuning and reinforcement learning objectives, allowing positive and negative offline completions, reference responses, and current and historical rollouts to share a common reward‑labeled format and replay buffer. Experiments on two model scales and four mathematical reasoning benchmarks show that CBPO, especially the CBPO‑Unified variant, outperforms baseline methods in avg@32 and pass@32 metrics, with historical rollouts providing additional gains.
arXiv:2509. 03206v2 Announce Type: replace-cross Abstract: Learning from reward functions and imitation learning of demonstrations are the two principal approaches for training autonomous systems that interact with an environment through action and observation.
The paper identifies an informational curse of horizon in goal-conditioned policy learning, showing that longer goal relabeling horizons during training can hurt policy generalization and performance. Controlled experiments with oracle planners separate training horizons from test horizons, revealing that behavioral cloning policies degrade with longer horizons while reinforcement learning mitigates this effect. The authors attribute this to reduced conditional mutual information between actions and hindsight-relabeled goals, and demonstrate that distilling short-horizon policy Jacobians into long-horizon policies improves performance, especially in combinatorial manipulation tasks.
arXiv:2407. 21359v2 Announce Type: replace-cross Abstract: Imagining potential outcomes of actions before execution helps agents make more informed decisions, a prospective thinking ability fundamental to human cognition.
arXiv:2610.15691v1 Announce Type: new Abstract: Existing reinforcement learning (RL) largely focuses on optimizing learning algorithms within a pre-defined task formulation, while applying RL to new...
MInTRL (Minimal Intervention Reinforcement Learning) expands exploration in on-policy reinforcement learning by inserting sparse, local corrections into rollouts via a judge-intervention policy. These interventions replace erroneous suffixes and immediately return control to the main policy, allowing the agent to explore beyond its natural trajectory while maintaining on-policy data. The method uses a sequence-level advantage-regression objective, avoiding importance sampling, and demonstrates superior performance on math and code benchmarks compared to standard on-policy and off-policy baselines.
arXiv:2610.19783v1 Announce Type: new Abstract: Reinforcement learning (RL) holds the promise of learning a policy that can outperform what is shown in the data. However, in practice, such systems su...
The paper introduces DCRL (Divide-and-Conquer RL), a method that recursively decomposes offline goal-conditioned reinforcement learning trajectories into a balanced binary tree. By training values from the leaves up to the root, DCRL avoids noisy max-based backups and reduces bootstrap depth from linear to logarithmic, thereby limiting error accumulation. Experiments on diverse goal-reaching tasks show that DCRL outperforms prior flat offline GCRL methods, achieving a higher average score on the most challenging long-horizon OGBench tasks.
The paper introduces Generalized Implicit Temporal Abstraction (GITA), a method for goal-conditioned reinforcement learning that conditions a single value function on multiple temporal abstraction levels (k). By aggregating advantage-weighted supervision across various k values, GITA preserves both long-range signal and local resolution without committing to a single k. Experiments on OGBench show that GITA outperforms existing offline GCRL baselines, improving average success rates by 25 percentage points over HIQL and 7 percentage points over OTA.