arXiv Machine Learning

Q-Learning with Scalar Adjoint Matching

The paper introduces Q‑Learning with Scalar Adjoint Matching (SQAM), a method that replaces expensive per‑step vector–Jacobian products in flow policy fine‑tuning with a closed‑form scalar adjoint that scales the value gradient by flow time. SQAM also incorporates a value penalty on policy‑generated actions, which is shown to be crucial for effective learning. Experiments demonstrate that SQAM outperforms strong baselines on the hardest OGBench domains and improves supervised fine‑tuning on a real bimanual robot with a vision‑language‑action policy.

arXiv Machine Learning

6d ago

QF3: Fast Flow RL with Filtered Q-Gradients

QF3 (Fast Flow RL with Filtered Q-Gradients) is an online off‑policy reinforcement learning algorithm that trains flow policies by combining flow matching with the critic’s action gradient, backpropagated through a one‑step prediction of the flow output. It selectively applies the critic gradient only to action dimensions close to the replay action to ensure reliable updates. QF3 is the first off‑policy flow RL method to train humanoid locomotion policies from scratch and transfer them zero‑shot to hardware, achieving a 10× wall‑clock speedup over the on‑policy FPO++ method, and it can also fine‑tune pretrained flow‑based manipulation policies on ABC‑Sim and Robomimic tasks.

By Chung Min Kim, Brent Yi, David McAllister, Hongsuk Choi, Himanshu Gaurav Singh, Jinkun Cao, Ken Goldberg, Pieter Abbeel, Carmelo Sferrazza, Angjoo Kanazawa

diffusion reinforcement-learning robotics fine-tuning efficiency

arXiv AI

Jun 10

Test-Time Gradient Guidance of Flow Policies in Reinforcement Learning

arXiv:2606. 11087v1 Announce Type: cross Abstract: Expressive continuous control policies, such as diffusion and flow models, form the backbone of recent advances in scaling imitation learning for simulated and real robot control.

By Zhiyuan Zhou, Andy Peng, Charles Xu, Qiyang Li, Tobias Springenberg, Kevin Frans, Sergey Levine

diffusion reinforcement-learning robotics benchmarks

Hugging Face Trending Papers

6d ago

QF3: Fast Flow RL with Filtered Q-Gradients

Flow policies have become a standard policy class for learning robot behaviors from demonstrations, but reinforcement learning is still critical for improving pre-trained flow policies or learning the...

diffusion reinforcement-learning robotics fine-tuning efficiency

arXiv AI

Jun 9

Reinforcement Learning for Flow-Matching Policies with Density Transport

arXiv:2606. 08602v1 Announce Type: cross Abstract: We present an online reinforcement learning (RL) algorithm for fine-tuning flow-matching policies in continuous-control problems.

By Boshu Lei, Kostas Daniilidis, Antonio Loquercio

diffusion reinforcement-learning robotics fine-tuning efficiency multimodal safety

arXiv AI

Jun 16

QPILOTS: Efficient Test-Time Q-Steering for Flow Policies

arXiv:2606. 14801v1 Announce Type: cross Abstract: Flow-matching and diffusion policies are expressive action generators, but optimizing them with temporal-difference reinforcement learning (RL) remains difficult.

By Yifan Ruan, Chenyang Cao, Andreas Burger, Ali Pesaranghader, Kaveh Kamali, Jaehong Kim, Nandita Vijaykumar, Alan Aspuru-Guzik, Igor Gilitschenski, Nicholas Rhinehart

diffusion reinforcement-learning robotics fine-tuning multimodal benchmarks

arXiv Machine Learning

Jun 26

Reinforcement Fine-Tuning of Flow-Matching Policies for Vision-Language-Action Models

arXiv:2510. 09976v2 Announce Type: replace Abstract: Vision-Language-Action (VLA) models such as OpenVLA, Octo, and $\pi_0$ have shown strong generalization by leveraging large-scale demonstrations, yet their performance is still fundamentally constrained by the quality and coverage of supervised data.

By Mingyang Lyu, Yinqian Sun, Erliang Lin, Huangrui Li, Ruolin Chen, Feifei Zhao, Yi Zeng

diffusion reinforcement-learning fine-tuning multimodal benchmarks

arXiv AI

Jul 14

VINE: Taming Generative Control Policies for Reinforcement Learning

arXiv:2607. 10369v1 Announce Type: cross Abstract: Flow-matching policies have emerged as an effective policy parameterization for robot learning.

By Rushuai Yang, Zhuo Han, Houlin Li, Hecheng Wang, Zhichao Wu, Rui Zhang, Zhaowei Zhang, Zihong Chen, Xiaohan Yan, Chiming Liu, Yi Chen, Wei Shan, Maoqing Yao

reinforcement-learning robotics multimodal benchmarks

arXiv AI

4d ago

Trust Region Q Adjoint Matching

arXiv:2605.27079v2 Announce Type: replace-cross Abstract: Off-policy reinforcement learning of pretrained flow policies remains challenging due to the instability of optimization arising from the mul...

By Yonghoon Dong, Kyungmin Lee, Changyeon Kim, Jaehyuk Kim, Jinwoo Shin

reinforcement-learning fine-tuning

arXiv AI

Jul 3

Guided Action Flow: Q-Guided Inference for Flow-Matching Vision-Language-Action Policies

arXiv:2607. 02092v1 Announce Type: cross Abstract: Flow-matching vision-language-action policies generate robot action chunks through an iterative transport process, creating an opportunity for test-time guidance without retraining the base policy.

By Liuhaichen Yang, Zhuang Jiang, Chenchao Sheng, Zezhi Tang

robotics multimodal

arXiv AI

Oct 2

DriftOPD: Sequence-Level Reverse-KL Distillation for One-Step VLA Policies

DriftOPD is a teacher‑free, rollout‑free framework that performs sequence‑level on‑policy distillation of continuous Vision‑Language‑Action (VLA) action experts. It decomposes the sequence‑level reverse‑KL divergence into a chunk‑level reverse‑KL term and a future‑potential term, optimizing them with a one‑step drifting objective and a Q‑function critic learned from offline demonstrations. Experiments on multiple VLA architectures in simulation and real‑world manipulation show that DriftOPD outperforms existing one‑step distillation baselines while matching the task success of multi‑step teacher policies.

By Youngjun Jun, Kyumin Choi, Youngmin Kim, Seonghyun Jin, Sunwoo Park, Jangho Park, Jong Chul Ye

reinforcement-learning robotics efficiency multimodal

arXiv AI

Aug 19

Q-Learning With World Models

The paper introduces QWM, a framework that integrates world models with standard Q‑learning to perform test‑time search over imagined trajectories. By training the policy and value function solely on real transitions, QWM avoids compounding model bias while still benefiting from predictive search. Experiments on the Robomimic and LIBERO manipulation benchmarks show that QWM outperforms strong prior state‑of‑the‑art methods in both sample efficiency and performance.

By Perry Dong, Yueru Jia, Chelsea Finn, Dorsa Sadigh

reinforcement-learning robotics fine-tuning multimodal benchmarks safety

arXiv Machine Learning

Aug 24

Decoupling Policy Extraction for Offline Reinforcement Learning

arXiv:2608.20909v1 Announce Type: new Abstract: Offline RL methods commonly jointly train the actor and critic, where the critic is used to guide the actor toward higher-value actions. This coupled l...

By Xuyao Lin, Yixiang Shan, Jinru Duan, Tao Yang, Xinyu Zhao, Runyu Lei, Yiming Zhao, Jiaxin Fan, Zongbao Feng, Peng Jia

rag reinforcement-learning

Related stories