Yichen Guo

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.

MiCo: Mutual Information Coverage Optimization through Semantic Erasure Modeling for Efficient MLLM Inference

Tinghao Wang, Yichen Guo, Qizhe Zhang, Yuan Zhang, Weimin Ouyang, Rui Huang, Jiajun Cao, Sixiang Chen, Hao Jiang, Jixian Wu, Zheng Lu, Bofan Zhu, Renyuan Li, Shanghang Zhang

Preprint · 2026 · Co-first author

Training-free, two-stage visual token pruning that maximizes mutual information coverage: select a representative visual pool, then refine it with task-aware signals.

ReMAP: Restoring the Perceptual Cycle with Reasoning-Time Latent Visual Memory

Hao Jiang*, Zhanyu Guo*, Chenwei Wu†, Yichen Guo†, Qizhe Zhang, Junchi Yao, Jixian Wu, Jinhao You, Kai Tang, Jiajun Cao, Tinghao Wang, Mengyu Wang, Leo Anthony Celi, Shanghang Zhang

* Equal contribution · † Project leaders

Preprint · 2026 · Project Leader

Global and local latent visual memories restore visual grounding during reasoning; a reinforcement-learning policy decides when to retrieve each type of evidence.

My contribution: Project leader, working on latent visual memory and learned memory-access policies for multimodal reasoning.

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.

Dataset access is currently restricted; data upload is pending.

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

NeurIPS 2026 · Accepted · Co-author

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

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.