#yingyan
LightNews LightNews
csro-bot.bsky.social
Mohammed Hassan, Zhenyang Chen, Zheng Wang, Zhixin Zhu, Tianlong Chen, Yingyan, Lin, Chaojian Li: T3R: Training-Free Two-Stage Token Refinement Towards Efficient and Robust VLA Models https://arxiv.org/abs/2611.08413 https://arxiv.org/pdf/2611.08413 https://arxiv.org/html/2611.08413
November 6, 2026 at 6:45 AM
cslg-bot.bsky.social
Avranshu Mahanty, Ran Jin, Yingyan Zeng: Class-aware Latent Attention Replay (CLAR): Class Incremental Hazard Diagnosis for Resilient AI in the Manufacturing Industrial Internet https://arxiv.org/abs/2610.39323 https://arxiv.org/pdf/2610.39323 https://arxiv.org/html/2610.39323
October 30, 2026 at 6:44 AM
csdl-bot.bsky.social
Yingyan Zeng, J. P. Panda, Avranshu Mahanty, Han Hu, Depeng Xu, Ying Sun: MultiphaseHUB for Evidence-Grounded Metadata Curation and Task-Aware Compression of Multiphase Data https://arxiv.org/abs/2610.28644 https://arxiv.org/pdf/2610.28644 https://arxiv.org/html/2610.28644
October 22, 2026 at 6:40 AM
news.karthihegde.dev
MultiphaseHUB for Evidence-Grounded Metadata Curation and Task-Aware Compression of Multiphase Data
tech_blogs_arxiv | Author: Yingyan Zeng, J. P. Panda, Avranshu Mahanty, Han Hu, Depeng Xu, Ying Sun

#DataEngineering
MultiphaseHUB for Evidence-Grounded Metadata Curation and Task-Aware Compression of Multiphase Data
Multiphase transport experiments and simulations generate high-dimensional data whose scientific value depends on the physical conditions, sensing configuration, calibration, synchronization, uncertainty, and provenance accompanying the raw signals. These data are difficult to reuse when experimenta
arxiv.org
October 22, 2026 at 4:06 AM
csai-bot.bsky.social
Chung-En Ho, Weiyu Sun, Cheng-Jhih Shih, He Li, Yong Liu, Yingyan Celine Lin: SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language Models https://arxiv.org/abs/2610.04875 https://arxiv.org/pdf/2610.04875 https://arxiv.org/html/2610.04875
October 7, 2026 at 6:40 AM
csai-bot.bsky.social
Chung-En Ho, Weiyu Sun, Cheng-Jhih Shih, He Li, Yong Liu, Yingyan, Lin: SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language Models https://arxiv.org/abs/2610.04875 https://arxiv.org/pdf/2610.04875 https://arxiv.org/html/2610.04875
October 6, 2026 at 6:40 AM
arxiv-cs-cv.bsky.social
Xianchi Dong, Yingyan Hou, Chao Ren, Wanxuan Lu, Zihan Wei, Hongfeng Yu, Yixiao Wang, Chubo Deng, Xian Sun
PDA++: Field-Aligned Planning and Scene-Adaptive Insertion in Remote Sensing
https://arxiv.org/abs/2609.18329
September 19, 2026 at 7:51 AM
quant-hub.bsky.social
🚨 $LXEH
Liang Yingyan () filed Form 3 reporting no initial stock ownership.
SEC Filing: LXEH
Click to view the official SEC document.
www.sec.gov
August 18, 2026 at 10:15 AM
paper.bsky.social
[30/30] 181 Likes, 4 Comments, 2 Posts
2604.18486, cs․CV | cs․CL | cs․RO, 08 May 2026

🆕Xiaomi OneVL: One-Step Latent Reasoning and Planning with Vision-Language Explanation

Jinghui Lu, Jiayi Guan, Zhijian Huang, Jinlong Li, Guang Li, Lingdong Kong, Yingyan Li, Han Wang, Shaoqing Xu, Yuechen...
Chain-of-Thought (CoT) reasoning has become a powerful driver of trajectory prediction in VLA-based autonomous driving, yet its autoregressive nature imposes a latency cost that is prohibitive for real-time deployment. Latent CoT methods attempt to close this gap by compressing reasoning into continuous hidden states, but consistently fall short of their explicit counterparts. We suggest that this is due to purely linguistic latent representations compressing a symbolic abstraction of the world, rather than the causal dynamics that actually govern driving. Thus, we present OneVL (One-step latent reasoning and planning with Vision-Language explanations), a unified VLA and World Model framework that routes reasoning through compact latent tokens supervised by dual auxiliary decoders. Alongside a language decoder that reconstructs text CoT, we introduce a visual world model decoder that predicts future-frame tokens, forcing the latent space to internalize the causal dynamics of road geometry, agent motion, and environmental change. A three-stage training pipeline progressively aligns these latents with trajectory, language, and visual objectives, ensuring stable joint optimization. In inference, the auxiliary decoders are discarded, and all latent tokens are prefilled in a single parallel pass, matching the speed of answer-only prediction. Across four benchmarks, OneVL becomes the first latent CoT method to surpass explicit CoT, delivering superior accuracy at answer-only latency. These results show that with world model supervision, latent CoT produces more generalizable representations than verbose token-by-token reasoning. Code has been open-sourced to the community. Project Page: https://xiaomi-embodied-intelligence.github.io/OneVL
May 18, 2026 at 12:06 AM
paper.bsky.social
[18/30] 92 Likes, 1 Comments, 2 Posts
2506.19850, cs․CV | cs․RO, 24 Jun 2025

🆕Unified Vision-Language-Action Model

Yuqi Wang, Xinghang Li, Wenxuan Wang, Junbo Zhang, Yingyan Li, Yuntao Chen, Xinlong Wang, Zhaoxiang Zhang
Vision-language-action models (VLAs) have garnered significant attention for their potential in advancing robotic manipulation. However, previous approaches predominantly rely on the general comprehension capabilities of vision-language models (VLMs) to generate action signals, often overlooking the rich temporal and causal structure embedded in visual observations. In this paper, we present UniVLA, a unified and native multimodal VLA model that autoregressively models vision, language, and action signals as discrete token sequences. This formulation enables flexible multimodal tasks learning, particularly from large-scale video data. By incorporating world modeling during post-training, UniVLA captures causal dynamics from videos, facilitating effective transfer to downstream policy learning--especially for long-horizon tasks. Our approach sets new state-of-the-art results across several widely used simulation benchmarks, including CALVIN, LIBERO, and Simplenv-Bridge, significantly surpassing previous methods. For example, UniVLA achieves 95.5% average success rate on LIBERO benchmark, surpassing pi0-FAST's 85.5%. We further demonstrate its broad applicability on real-world ALOHA manipulation and autonomous driving.
July 4, 2025 at 12:09 AM
paper.bsky.social
[29/30] 185 Likes, 5 Comments, 4 Posts
2506.02153, cs․AI, 02 Jun 2025

🆕Small Language Models are the Future of Agentic AI

Peter Belcak, Greg Heinrich, Shizhe Diao, Yonggan Fu, Xin Dong, Saurav Muralidharan, Yingyan Celine Lin, Pavlo Molchanov
Large language models (LLMs) are often praised for exhibiting near-human performance on a wide range of tasks and valued for their ability to hold a general conversation. The rise of agentic AI systems is, however, ushering in a mass of applications in which language models perform a small number of specialized tasks repetitively and with little variation. Here we lay out the position that small language models (SLMs) are sufficiently powerful, inherently more suitable, and necessarily more economical for many invocations in agentic systems, and are therefore the future of agentic AI. Our argumentation is grounded in the current level of capabilities exhibited by SLMs, the common architectures of agentic systems, and the economy of LM deployment. We further argue that in situations where general-purpose conversational abilities are essential, heterogeneous agentic systems (i.e., agents invoking multiple different models) are the natural choice. We discuss the potential barriers for the adoption of SLMs in agentic systems and outline a general LLM-to-SLM agent conversion algorithm. Our position, formulated as a value statement, highlights the significance of the operational and economic impact even a partial shift from LLMs to SLMs is to have on the AI agent industry. We aim to stimulate the discussion on the effective use of AI resources and hope to advance the efforts to lower the costs of AI of the present day. Calling for both contributions to and critique of our position, we commit to publishing all such correspondence at https://research.nvidia.com/labs/lpr/slm-agents.
June 29, 2025 at 12:07 AM
paper.bsky.social
[25/30] 90 Likes, 4 Comments, 3 Posts
2505.04588, cs․CL, 07 May 2025

🆕ZeroSearch: Incentivize the Search Capability of LLMs without Searching

Hao Sun, Zile Qiao, Jiayan Guo, Xuanbo Fan, Yingyan Hou, Yong Jiang, Pengjun Xie, Fei Huang, Yan Zhang
Effective information searching is essential for enhancing the reasoning and generation capabilities of large language models (LLMs). Recent research has explored using reinforcement learning (RL) to improve LLMs' search capabilities by interacting with live search engines in real-world environments. While these approaches show promising results, they face two major challenges: (1) Uncontrolled Document Quality: The quality of documents returned by search engines is often unpredictable, introducing noise and instability into the training process. (2) Prohibitively High API Costs: RL training requires frequent rollouts, potentially involving hundreds of thousands of search requests, which incur substantial API expenses and severely constrain scalability. To address these challenges, we introduce ZeroSearch, a reinforcement learning framework that incentivizes the search capabilities of LLMs without interacting with real search engines. Our approach begins with lightweight supervised fine-tuning to transform the LLM into a retrieval module capable of generating both relevant and noisy documents in response to a query. During RL training, we employ a curriculum-based rollout strategy that incrementally degrades the quality of generated documents, progressively eliciting the model's reasoning ability by exposing it to increasingly challenging retrieval scenarios. Extensive experiments demonstrate that ZeroSearch effectively incentivizes the search capabilities of LLMs using a 3B LLM as the retrieval module. Remarkably, a 7B retrieval module achieves comparable performance to the real search engine, while a 14B retrieval module even surpasses it. Furthermore, it generalizes well across both base and instruction-tuned models of various parameter sizes and is compatible with a wide range of RL algorithms.
May 14, 2025 at 12:07 AM
arxiv-cs-ro.bsky.social
Xiaomi Robotics Team, Jun Guo, Piaopiao Jin, Jason Li, Peiyan Li, Yingyan Li, Futeng Liu, Wanli Peng, Optimus Qin, Yifei Su, Nan Sun, Qiao Sun, ...
Xiaomi-Robotics-1: Scaling Vision-Language-Action Models with over 100K Hours of Real-World Trajectories
https://arxiv.org/abs/2607.15330
July 20, 2026 at 6:40 AM
arxiv-cs-ro.bsky.social
Shuyao Shang, Yuntao Chen, Yuqi Wang, Yingyan Li, Zhaoxiang Zhang
DriveDPO: Policy Learning via Safety DPO For End-to-End Autonomous Driving
https://arxiv.org/abs/2509.17940
September 23, 2025 at 5:55 AM
arxiv-cs-ro.bsky.social
Xiaohe Li, Feilong Huang, Zide Fan, Fangli Mou, Yingyan Hou, Chen Qian, Lijie Wen
MetaTra: Meta-Learning for Generalized Trajectory Prediction in Unseen Domain
https://arxiv.org/abs/2402.08221
February 14, 2024 at 5:02 AM
csar-bot.bsky.social
Zhongzhi Yu, Mingjie Liu, Michael Zimmer, Yingyan Celine Lin, Yong Liu, Haoxing Ren: Spec2RTL-Agent: Automated Hardware Code Generation from Complex Specifications Using LLM Agent Systems https://arxiv.org/abs/2506.13905 https://arxiv.org/pdf/2506.13905 https://arxiv.org/html/2506.13905
June 18, 2025 at 6:54 AM
nber.org
Exit barriers are entry barriers and in India they drive the low share of manufacturing production, from Shoumitro Chatterjee, Kala Krishna, Kalyani Padmakumar, and Yingyan Zhao https://www.nber.org/papers/w33830
May 28, 2025 at 11:00 PM
csir-bot.bsky.social
Hao Sun, Zile Qiao, Bo Wang, Guoxin Chen, Yingyan Hou, Yong Jiang, Pengjun Xie, Fei Huang, Yan Zhang: DecoupleSearch: Decouple Planning and Search via Hierarchical Reward Modeling https://arxiv.org/abs/2510.21712 https://arxiv.org/pdf/2510.21712 https://arxiv.org/html/2510.21712
October 28, 2025 at 6:32 AM
epidancer.bsky.social
Infusing #MELODEM's zoom ether in a couple hours (11a US ET):

"How to assess #cognitivedecline when test administration changes across study waves? #Harmonizing cognitive scores across waves in the China Health & Retirement Longitudinal Study," by Yingyan Wu.

Thanks, NIA/NIH.
epidancer.bsky.social Jennifer Weuve @epidancer.bsky.social · Nov 21
Tmw (Th) 11a US ET on channel #MELODEM:

Yingyan Wu: "How to assess #cognitivedecline when test administration changes across study waves? #Harmonizing cognitive scores across waves in the China Health & Retirement Longitudinal Study"

Hop on our distrib list 4 info sites.bu.edu/melodem/cale...
Calendar | Methods for Longitudinal Studies in Dementia (MELODEM)
sites.bu.edu
November 21, 2024 at 1:36 PM
epidancer.bsky.social
Tmw (Th) 11a US ET on channel #MELODEM:

Yingyan Wu: "How to assess #cognitivedecline when test administration changes across study waves? #Harmonizing cognitive scores across waves in the China Health & Retirement Longitudinal Study"

Hop on our distrib list 4 info sites.bu.edu/melodem/cale...
Calendar | Methods for Longitudinal Studies in Dementia (MELODEM)
sites.bu.edu
November 21, 2024 at 3:41 AM
cscv-bot.bsky.social
Mengqi Shen, Haicheng Wang, Meghna Trivedi, Tony J. Wang, Yuanguang Xu, Yingyan Zeng, Yading Yuan: Anatomically Consistent Cross-Contrast Super-Resolution of Anisotropic Brain T2w MRI https://arxiv.org/abs/2608.08401 https://arxiv.org/pdf/2608.08401 https://arxiv.org/html/2608.08401
August 11, 2026 at 6:41 AM
cscv-bot.bsky.social
Yufeng Hong, Xiaotian Zhou, Yingyan Li, Xiangpo Zhou, Lin Liu, Yadan Luo, Shaoqing Xu, Lei Yang, Ziying Song: DriveFuture: Future-Aware Latent World Models for Autonomous Driving https://arxiv.org/abs/2605.09701 https://arxiv.org/pdf/2605.09701 https://arxiv.org/html/2605.09701
May 12, 2026 at 6:43 AM
cscv-bot.bsky.social
Yingyan Xu, Pramod Rao, Sebastian Weiss, Gaspard Zoss, Markus Gross, Christian Theobalt, Marc Habermann, Derek Bradley: RelightAnyone: A Generalized Relightable 3D Gaussian Head Model https://arxiv.org/abs/2601.03357 https://arxiv.org/pdf/2601.03357 https://arxiv.org/html/2601.03357
January 8, 2026 at 6:30 AM
cscv-bot.bsky.social
Zhenbang Du, Yonggan Fu, Lifu Wang, Jiayi Qian, Xiao Luo, Yingyan, Lin: Fewer Denoising Steps or Cheaper Per-Step Inference: Towards Compute-Optimal Diffusion Model Deployment https://arxiv.org/abs/2508.06160 https://arxiv.org/pdf/2508.06160 https://arxiv.org/html/2508.06160
August 11, 2025 at 6:31 AM