Less is More: Reducing Complexity in Vision-Language-Action Systems

Jinhui Ye^*, Ning Gao^*, Senqiao Yang, Jinliang Zheng, Zixuan Wang, Yuxin Chen, Pengguang Chen, Yilun Chen, Shu Liu, Jiaya Jia

A simple VLA baseline for studying action modeling, robot pretraining, and interface design under controlled conditions.

Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments

Qiuyue Wang, Mingsheng Li, Jian Guan, Jinhui Ye, Sicheng Xie, Yitao Liu, et al.

A generalist policy across robot embodiments and tasks. My work focused on training and evaluation infrastructure.

ST4VLA: Spatially Guided Training for Vision-Language-Action Model

Jinhui Ye^*, Fangjing Wang^*, Ning Gao^*, Junqiu Yu^*, Yangkun Zhu, Bin Wang, Jinyu Zhang, Weiyang Jin, Yanwei Fu, Feng Zheng, Yilun Chen, Jiangmiao Pang

Spatially guided training connects VLM grounding with action learning. The ICLR version of InternVLA-M1.

StarVLA: A Lego-like Codebase for Vision-Language-Action Model Developing

StarVLA Community

Composable infrastructure for VLA research. Project founder, core architect, and lead maintainer.

Re-thinking Temporal Search for Long-Form Video Understanding

Jinhui Ye^*, Zihan Wang^*, Haosen Sun, Keshigeyan Chandrasegaran, Zane Durante, Cristobal Eyzaguirre, Yonatan Bisk, Juan Carlos Niebles, Ehsan Adeli, Li Fei-Fei, Jiajun Wu, Manling Li

Find the few relevant frames among tens of thousands before performing detailed video reasoning.

MolErr2Fix: Benchmarking LLM Trustworthiness in Chemistry via Modular Error Detection, Localization, Explanation, and Revision

Yuyang Wu^*, Jinhui Ye^*, Shuhao Zhang^*, Lu Dai, Yonatan Bisk, Olexandr Isayev

A benchmark for detecting, locating, explaining, and revising chemistry errors in language-model outputs.

FACE: A General Framework for Mapping Collaborative Filtering Embeddings into LLM Tokens

Chao Wang^*, Yixin Song^*, Jinhui Ye, Chuan Qin, Dazhong Shen, Lingfeng Liu, Xiang Wang, Yanyong Zhang

Mapping collaborative filtering representations into language-model tokens for interpretable recommendation.

SePer: Measure Retrieval Utility Through the Lens of Semantic Perplexity Reduction

Lu Dai, Yijie Xu, Jinhui Ye, Hao Liu, Hui Xiong

Measure retrieval utility by how much the evidence reduces uncertainty in a language model's answer.

Improving Gloss-free Sign Language Translation by Reducing Representation Density

Jinhui Ye, Xing Wang, Wenxiang Jiao, Junwei Liang, Hui Xiong

Contrastive learning for more discriminative visual representations in gloss-free sign language translation.

Cross-modality Data Augmentation for End-to-End Sign Language Translation

Jinhui Ye, Wenxiang Jiao, Xing Wang, Zhaopeng Tu, Hui Xiong

Transfer gloss-to-text knowledge to video-to-text translation through cross-modality data augmentation.

Task illustration: PHOENIX14T · Camgoz et al., CVPR 2018

Scaling Back-Translation with Domain Text Generation for Sign Language Gloss Translation

Jinhui Ye^*, Wenxiang Jiao^*, Xing Wang, Zhaopeng Tu

Generate in-domain spoken-language text to scale back-translation when parallel sign-language data is scarce.