2605.27079v1 May 26, 2026 cs.LG

Trust Region Q Adjoint Matching

Changyeon Kim
Changyeon Kim
Citations: 289
h-index: 9
Kyungmin Lee
Kyungmin Lee
Citations: 317
h-index: 9
Yong Dong
Yong Dong
Citations: 7
h-index: 2
Jinwoo Shin
Jinwoo Shin
Citations: 876
h-index: 10
Jaehyuk Kim
Jaehyuk Kim
Citations: 78
h-index: 2

Off-policy reinforcement learning of pretrained flow policies remains challenging due to the instability of optimization arising from the multi-step sampling process. Recently, Q-learning with Adjoint Matching (QAM) addressed this issue by reformulating into a memoryless stochastic optimal control (SOC) problem with a learned critic. However, QAM inherits a fundamental fragility of critic-guided improvement: small critic errors are amplified when critics are ill-conditioned, often leading to model collapse. This paper introduces Trust Region Q-Adjoint Matching (TRQAM), a stable off-policy fine-tuning algorithm that adaptively controls the path-space KL with pretrained flow policies through projected dual descent. Specifically, we optimize the trust-region parameter $λ$ in SOC dynamics, and theoretically show that the path-space KL can be represented by a closed-form function of $λ$. As a result, our method can precisely control the exact deviation from pretrained flow policies, achieving stable off-policy RL. Through experiments on 50 OGBench tasks, TRQAM consistently outperforms prior arts in both offline RL and offline-to-online RL. In particular, TRQAM achieves an overall success rate of 68% in offline RL, substantially improves the strongest baseline at 46%.

1 Citations
0 Influential
5 Altmetric
26.0 Score
Original PDF

No Analysis Report Yet

This paper hasn't been analyzed by Gemini yet.

Log in to request an AI analysis.

댓글

댓글을 작성하려면 로그인하세요.

아직 댓글이 없습니다. 첫 번째 댓글을 남겨보세요!