8 papers indexed
Haisheng Su, Zongdai Liu, Xin Jin, Haoxuan Dou, Chengming Hu, Baorun Li, et al.
World Action Models (WAMs) offer a promising paradigm for robotic manipulation by jointly modeling visual state transitions and robot actions. However, existing WAMs are constrained by limited temporal context, coarse episode-level language supervision, and predominantly text-onl…
Xiangdong Zhang, Xiaohan Qin, Sunan Zou, Tuo Dai, Xiaoming Shi, Huaijin Wu, et al.
Hyper-Connections (HC) expand the residual stream of Transformers into $N$ parallel streams, providing a form of memory scaling beyond model width and depth. Manifold-Constrained HC (mHC) stabilizes this formulation at scale. The large gains from $N{=}1$ to $N{=}4$ suggest residu…
Jiatong Zhao, Tengyue Zhang, Yuhan Wang, Fuyuan Wu, Junchi Yan
Offline-trained surrogates for Inertial Confinement Fusion (ICF) suffer a well-known failure mode that iterative optimizers drive inputs into out-of-distribution (OOD) regions where predictions become unreliable. Here we present Co4ICF, a co-evolving framework that couples a phys…
Songbur Wong, Xiaosong Jia, Junqi You, Bo Zhang, Pei Xu, Renqiu Xia, et al.
Evaluating end-to-end autonomous driving (E2E-AD) remains challenging, as existing driving simulation methods often trade off closed-loop interactivity (e.g., CARLA) and real-world visual fidelity (e.g., nuScenes). We present \textbf{\emph{Point as Skeleton}}, a generative sensor…
ACE-Brain Team, :, Ziyang Gong, Haoming Gu, Zehang Luo, Tianyi Zhang, et al.
Embodied AI is moving from isolated perception or action modules toward physical agents that understand, plan under goals, act through robot bodies, monitor progress, and improve from experience. Existing systems address this loop only in parts: end-to-end policies generate actio…
Zhaokai Wang, Mingxin Liu, Zirun Zhu, Ziqian Fan, Yiguo He, Mohan Zhang, et al.
Recent image generation and editing models can produce visually appealing natural images, yet they remain unreliable when the target image is a knowledge-intensive diagram whose correctness depends on disciplinary concepts, symbolic structure, and precise spatial relations. We in…
Ning Liao, Zihao Long, Xiaoxing Wang, Xue Yang, Yaoming Wang, Ziyuan Zhuang, et al.
The increasing complexity of agentic tasks has led to rapidly growing trajectory lengths, which poses significant challenges for large language model (LLM) based agents with fixed context windows. Existing context management techniques, such as truncation and summarization, suffe…
Tao Lin, Yuxin Du, Yiran Mao, Zewei Ye, Yilei Zhong, Bing Cheng, et al.
Vision-Language-Action (VLA) models are commonly pretrained on robot demonstrations by jointly mapping visual observations and language instructions to actions. However, dense visual-action supervision can dominate the comparatively sparse language-action signal. As a result, pol…