From Pixels to Actions - Towards Native Vision-Language-Action Policies

weichenfan, Haiwen Diao, Penghao Wu, Siyuan Hu, Ziqi Huang, Tianqi Liu, Mingze Sun, Ziyi Wang, Zhongang Cai, Hanming Deng, Quan Wang, yl-1993, Lewei Lu, Fangzhou Hong, Zhaoxi Chen, Dahua Lin, Ziwei Liu

From Pixels to Actions - Towards Native Vision-Language-Action Policies: 60 upvotes on Hugging Face Daily Papers, #2 of 38 papers on 2026-11-05. Day-by-day upvote history.

Upvotes over time

It had 56 upvotes by its first day, 58 upvotes by the next day, 61 upvotes by day 3, 60 upvotes by day 7, and has 60 now. It ranked #16 of 207 papers in its week and #23 of 340 in its month.

Vision-language-action models are typically built on a simple premise: perception is first compressed by visual encoders, and action reasoning begins only afterwards. Yet manipulation often hinges on precisely the visual signals most vulnerable to this abstraction, e.g., local geometry, boundaries, depth, and subtle state changes. This raises a fundamental question: should action policies inherit a fixed visual endpoint, or directly access evolving visual computation? We introduce NEO-VLA, a structurally encoder-free vision-language-action framework. Images are tokenized by lightweight patch convolutional layers and evolve jointly with language throughout the backbone, allowing a flow-matching action expert to draw from instruction-conditioned visual representations at different depths. This simple change reveals a consistent pattern. Native architectures outperform encoder-based counterparts under controlled comparisons; full fine-tuning reduces but does not erase the gap, and additional scale alone cannot recover it. More strikingly, layer-wise probing shows that the action-critical signals often peak well before the final visual representation. Across diverse manipulation settings and distribution shifts, NEO-VLA consequently achieves stronger and more robust control. Together, these results suggest that embodied control benefits from access to visual representations across the computational hierarchy, rather than a fixed encoder output.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.