~ similar to 2606.26006· 20 results
Tongyan Fang, Siyuan Huang, Naiyu Fang, Ganlong Zhao +5 more
The paper proposes Hierarchical Advantage-Weighted Behavior Cloning (HABC) to improve online reinforcement learning by addressing issues with existing approaches.
Christian Scherer, Joe Watson, Theo Gruner, Daniel Palenicek +2 more
The paper proposes a coherent inverse reinforcement learning (IRL) method to improve large behavior models for robotic control, achieving superior sample efficiency and performance on complex sparse m…
This paper studies the effectiveness of pretraining the Q-function in value-based reinforce learning and finds that naive Q-function pretraining often provides little benefit over random initializatio…
The paper proposes an active, continual learning paradigm for Vision-Language-Action models to improve fine-tuning efficiency.
This paper introduces Act2Answer, a protocol to evaluate embodied vision-language-action models by requiring them to answer through action.
The paper proposes CTRL-STEER, a closed-loop framework that adaptively adjusts intervention strength to stabilize concept regulation and improve task success in Vision-Language-Action models without r…
Yuhao Wu, Yitian Liu, Weijie Shen, Mishuo Han +12 more
The paper introduces dVLA-RL, a method for training Vision-Language-Action models using Reinforcement Learning.
Zhongxi Chen, Yifan Han, Yanming Shao, Huanming Liu +4 more
BORA is an offline-to-online RL framework that enhances dexterous VLA models for real-world robotics by using an action-conditioned critic and a lightweight residual adaptation mechanism to correct ex…
Dong Jing, Jingchen Nie, Tianqi Zhang, Jiaqi Liu +3 more
TempoVLA is a novel Vision-Language-Action model that enables controllable execution speed for robot manipulation by explicitly conditioning the policy on the desired speed.
Qi Liu, Mingdi Sun, Yongyi He, Zhi Zheng +4 more
The paper proposes EKSFT, a selective fine-tuning method that masks high-entropy or high-KL divergence tokens during Supervised Fine-Tuning (SFT) to prevent distribution shift and improve subsequent R…
Mingjian Gao, Wenqiao Zhang, Yuqian Yuan, Yang Dai +8 more
VISUALTHINK-VLA introduces a visual intermediate-reasoning framework that guides action prediction using compact visual evidence, achieving high accuracy and significantly low latency for real-time Vi…
This paper proposes a reinforcement learning framework for Computer-Use Agents that uses autonomous vision-language evaluation as a scalable supervision signal.
Jingdong Zhang, Jionghao Wang, Zhengming Yu, Yancheng Cai +18 more
The paper introduces Flex-RSI, a benchmark for evaluating the recursive self-improvement of agents on vision and robotics tasks.
Haofei Yu, Yixing Chen, Yining Zhao, Xuyan Ye +5 more
The paper proposes a new method, ame, to improve the performance of language agents in simulated and realistic environments by using multi-step action targets.
This paper presents a policy that combines a Human300-fine-tuned GR00T-N1.5 action generator, an action-conditioned graph dynamics model, and a learned binary state-reward scorer to improve robotic ta…
Zizhe Chen, Jiqian Dong, Yizhou Tian, Garry Yang +3 more
This paper introduces Numca and Hista, two novel techniques that significantly improve state value estimation for LLM reinforcement learning, addressing the instability of standard critic approaches.
The paper proposes NestedVLA, a paradigm for consolidating skills learned across heterogeneous tasks and environments by generating task-adaptive action-expert parameters.
Tong Liu, Cheng Qian, Matej Cief, Yuan He +3 more
This paper analyzes tool-calling in LLM agents, demonstrating that evaluation results are highly sensitive to implementation details and proposing new techniques to significantly improve the efficienc…