paper-with-me

홈 › Papers

Balancing Signal and Variance: Adaptive Offline RL Post-Training for VLA Flow Models

2025-09-04 · Hongyin Zhang, Shiyuan Zhang, Junxi Jin, Qixin Zeng, Yifan Qiao, Hongchao Lu, Donglin Wang arxiv

Vision-Language-Action (VLA) models based on flow matching have shown excellent performance in general-purpose robotic manipulation tasks. However, the action accuracy of these models on complex downstream tasks is unsatisfactory. One important reason is that these models rely solely on the post-training paradigm of imitation learning, which makes it difficult to have a deeper understanding of the distribution properties of data quality, which is exactly what Reinforcement Learning (RL) excels at. In this paper, we theoretically propose an offline RL post-training objective for VLA flow models and induce an efficient and feasible offline RL fine-tuning algorithm -- Adaptive Reinforced Flow Matching (ARFM). By introducing an adaptively adjusted scaling factor in the VLA flow model loss, we construct a principled bias-variance trade-off objective function to optimally control the impact of RL signal on flow loss. ARFM adaptively balances RL advantage preservation and flow loss gradient variance control, resulting in a more stable and efficient fine-tuning process. Extensive simulation and real-world experimental results show that ARFM exhibits excellent generalization, robustness, few-shot learning, and continuous learning performance.

📄 PDF Abstract BibTeX arXiv:2509.04063

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningFew-Shot LearningOffline RL

Similar Papers 제목 키워드 기반

Cross-Epoch Adaptive Rollout Optimization for RL Post-Training

2026-06-04 · Yiming Zong, Yige Wang, Jiashuo Jiang arxiv

LLM post-training often relies on reinforcement learning methods that sample multiple rollouts per prompt, yet most existing approaches use a fixed rollout budget for every prompt, despite large differences in the traini…

Reinforcement Learning

Adversarial Fine-tuning in Offline-to-Online Reinforcement Learning for Robust Robot Control

2025-10-15 · Shingo Ayabe, Hiroshi Kera, Kazuhiko Kawamoto arxiv

Offline reinforcement learning enables sample-efficient policy acquisition without risky online interaction, yet policies trained on static datasets remain brittle under action-space perturbations such as actuator faults…

Reinforcement Learning

MAPLE: Modality-Aware Post-training and Learning Ecosystem

2026-02-12 · Nikhil Verma, Minjung Kim, JooYoung Yoo, Kyung-Min Jin 외 arxiv

Multimodal language models now integrate text, audio, and video for unified reasoning. Yet existing RL post-training pipelines treat all input signals as equally relevant, ignoring which modalities each task actually req…

Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation

2026-06-30 · Jiachun Li, David Simchi-Levi arxiv

Adaptive experiments for average treatment effects (ATE) require randomized allocations balancing valid inference with statistical efficiency. The oracle design is a covariate-dependent Neyman rule governed by unknown ar…

GAC: Noise-Aware Adaptive Mixing for Hybrid SFT-RL Post-Training

2026-05-25 · Yuelin Hu, Zhenbo Yu, Zhengxue Cheng, Wei Liu 외 arxiv

Hybrid post-training usually combines supervised fine-tuning and reinforcement learning, but fixed mixing schedules cannot adapt when the relative noise of the two signals changes over time. We propose GAC, a noise-aware…

Reinforcement Learning