11 papers
LoopVL: Recurrent Visual Intelligence
Zhe Qian, Ziyang Gong, Zhongxing Xu +13
We introduce LoopVL to study whether Loop Transformers can be effectively extended to vision- language models. LoopVL combines Module-Loop and Model-Loop computation to iteratively…
When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression
Mingxuan Wang, Fei Luo, Bo Wang +6
Long horizon language model agents continually accumulate reasoning history, increasing context length and inference cost even after earlier decisions have been executed and observ…
Stable Geometry with Divergent Task Evidence for Efficient Long-Horizon Agent Compression
Mingxuan Wang, Fei Luo, Bo Wang +6
Long horizon agents accumulate growing interaction histories that increase context and inference costs. We find that geometric redundancy alone is an insufficient criterion for saf…
StateComp: Learning When to Compress History in Long Horizon Agents
Mingxuan Wang, Hongyue Chen, Yinglong Guo +6
Long-horizon agents continuously accumulate interaction history during task execution, yet the importance of past interactions changes as the agent state evolves. Existing context…
Memory Control Signals Emerge Before Action in Long Horizon Agents
Mingxuan Wang, Guorun Yao, Fei Luo +6
Long horizon language model agents continuously accumulate interaction history, increasing computational cost while making relevant information harder to preserve and reuse. Existi…
DRSR: Learning Set-Level Deletion Risk for Efficient Long-Horizon Agents
Mingxuan Wang, Bo Wang, Fei Luo +6
Long-horizon language-model agents accumulate reasoning traces, tool exchanges, and observations whose relevance changes with the current decision. Existing compression strategies…
SVSR: A Self-Verification and Self-Rectification Paradigm for Multimodal Reasoning
Zhe Qian, Nianbing Su, Zhonghua Wang +6
Current multimodal models often suffer from shallow reasoning, leading to errors caused by incomplete or inconsistent thought processes. To address this limitation, we propose Self…
Cognitive Pivot Points and Visual Anchoring: Unveiling and Rectifying Hallucinations in Multimodal Reasoning Models
Zhe Qian, Yanbiao Ma, Zhuohan Ouyang +7
Multimodal Large Reasoning Models (MLRMs) have achieved remarkable strides in visual reasoning through test time compute scaling, yet long chain reasoning remains prone to hallucin…
Reasoning emerges from constrained inference manifolds in large language models
Yanbiao Ma, Fei Luo, Linfeng Zhang +10
Reasoning in large language models is predominantly evaluated through labeled benchmarks, conflating task performance with the quality of internal inference. Here we study reasonin…
Geometric Prior-Guided Federated Prompt Calibration
Fei Luo, Ziwei Zhao, Mingxuan Wang +5
Federated Prompt Learning (FPL) offers a parameter-efficient solution for collaboratively training large models, but its performance is severely hindered by data heterogeneity, whi…
From Perception to Cognition: A Survey of Vision-Language Interactive Reasoning in Multimodal Large Language Models
Chenyue Zhou, Mingxuan Wang, Yanbiao Ma +19
Multimodal Large Language Models (MLLMs) strive to achieve a profound, human-like understanding of and interaction with the physical world, but often exhibit a shallow and incohere…