Publications · Yichen Guo

STAR-Pro's Adaptive Stage preserves visual feature coverage, followed by progressive token refinement across decoder layers.

STAR-Pro: Stage-Wise Token Adaptive Reduction with Progressive Refinement for Efficient Large Vision-Language Models

Yichen Guo, Tinghao Wang, Qizhe Zhang, Lingbei Meng, Yuan Zhang, Jiajun Cao, Hao Jiang, Chenwei Wu, Jixian Wu, Sixiang Chen, Tao Luo, Hongyang Cheng, Kai Tang, Chenxi Li, Renyuan Li, Xiande Huang, Wenya Wang, Shanghang Zhang

Under review · 2026 · Co-first author

Training-free visual token reduction that preserves feature coverage, then progressively refines retained tokens as language and vision interact.

My contribution: As a co-first author, I contributed to the design of adaptive-to-progressive token pruning and its evaluation across vision-language architectures.

Paper Code Project page

WorldEcho and WorldSync: diagnosing action following, aligning world models, and validating policy learning.

Do Robotic World Models Really Follow Actions? Diagnosing and Aligning Action-Conditioned Generation for Policy Learning

Sixiang Chen, Jiaming Liu, Jixian Wu, Yichen Guo, Tinghao Wang, Siyuan Qian, Hao Chen, Jiajun Cao, Jian Tang, Shanghang Zhang

Preprint · 2026 · Core contributor

WorldEcho diagnoses action-following failures; WorldSync aligns generation with robot dynamics to support policy learning.

My contribution: As a core contributor, I work on understanding and improving action following in robotic world models.

Paper CodeComing soon Project page

Hugging FacePrivate dataset

Dataset access is currently restricted; data upload is pending.

SafeDojo's pipeline combines action-conditioned world-model rollouts, task-success and safety evaluation, and constrained GRPO policy updates.

SafeDojo: Safe Reinforcement Learning for VLA via Interactive World Model

Kai Tang, Peidong Jia, Zhong Chu, Jixian Wu, Rui Ma, Jiajun Cao, Fangyuan Zhao, Sixiang Chen, Yichen Guo, Xiaowei Chi, Chun-Kai Fan, Kevin Zhang, Jinchang Xu, Fubing Yang, Weishi Mi, Xiaozhu Ju, Jian Tang, Shanghang Zhang

CoRL 2026 · Accepted · Co-author

Safe reinforcement learning for VLA policies through imagined rollouts in an interactive world model, balancing task progress and safety.

Paper CodeComing soon Project page

RegimeVGGT uses a layer-wise U-shaped schedule for token merging and spatially preserving key/value downsampling in visual geometry transformers.

RegimeVGGT: Layer-Wise Spatially Preserving Redundancy Removal for Visual Geometry Grounded Transformer

Jinhao You, Shuo Lyu, Zhuohang Lyu, Tanxuan Li, Zibo Zhao, Jiaxiang Hu, Kai Tang, Yichen Guo

Under review · 2026 · Co-author

Training-free acceleration of visual geometry transformers with layer-wise token merging and spatially preserving K/V downsampling.

Paper CodeComing soon Project page

SeeMe: Mitigating Hallucinations in Large Vision-Language Models through Effective Visual Token Engineering

J. You*, K. Tang*, Yichen Guo, C. Li, H. Li, Y. Qu, X. Huang

Manuscript · 2026 · Co-author

Visual token engineering to improve consistency and reduce hallucinations in vision-language models.

PaperComing soon CodeComing soon Project pageComing soon

MAP: Mitigating Hallucinations in Large Vision-Language Models with Map-Level Attention Processing

C. Li*, Yichen Guo*, B. Qian, J. You, K. Tang, Y. Du, Z. Zhang, X. Huang

Manuscript · 2026 · Equal contribution

Training-free decoding with map-level attention processing to mitigate hallucinations in vision-language models.

PaperComing soon CodeComing soon Project pageComing soon

STAR: Stage-Wise Attention-Guided Token Reduction for Efficient Large Vision-Language Models Inference

Yichen Guo, H. Li, Z. Zhang, J. You, K. Tang, X. Huang

Manuscript · 2025 · First author

Earlier STAR manuscript on two-stage, training-free visual token reduction; the current preprint is STAR-Pro.

PaperComing soon CodeComing soon Project pageComing soon

Mitigating Hallucinations via Inter-Layer Consistency Aggregation in Large Vision-Language Models

K. Tang*, J. You*, X. Ge, H. Li, Yichen Guo, X. Huang

Manuscript · 2025 · Co-author

Training-free decoding that aggregates inter-layer representations to improve semantic consistency and reduce hallucinations.

PaperComing soon CodeComing soon Project pageComing soon