11 papers
Relax Within, Balance Across: Geometry-Guided Load Balancing for Vision-Language Mixture-of-Experts
Ziang Wu, Peng Jin, Qishen Yin +4
Vision-language MoE batches contain different numbers of image and text tokens. Image resolution, image count, tiling, and prompt length all change this token mix. We call the stan…
V-Zero: Answer-Label-Free On-Policy Distillation with Contrastive Evidence Gating for Fine-Grained Visual Reasoning
Haoxiang Sun, Zhihang Yi, Langxuan Deng +6
Fine-grained visual reasoning requires multimodal large language models (MLLMs) to identify task-relevant visual evidence and ground their reasoning in local image regions. Existin…
AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin
Shuo Yang, Qihui Zhang, Yuyang Liu +7
Fine-tuning large language models (LLMs) improves performance but introduces critical safety vulnerabilities: even minimal harmful data can severely compromise safety measures. We…
Turning the TIDE: Cross-Architecture Distillation for Diffusion Large Language Models
Gongbo Zhang, Wen Wang, Ye Tian +1
Diffusion large language models (dLLMs) offer parallel decoding and bidirectional context, but state-of-the-art dLLMs require billions of parameters for competitive performance. Wh…
Benchmarking PhD-Level Coding in 3D Geometric Computer Vision
Wenyi Li, Renkai Luo, Yue Yu +5
AI-assisted coding has rapidly reshaped software practice and research workflows, yet today's models still struggle to produce correct code for complex 3D geometric vision. If mode…
Dr. Seg: Revisiting GRPO Training for Visual Large Language Models through Perception-Oriented Design
Haoxiang Sun, Tao Wang, Chenwei Tang +2
Following the success of Group Relative Policy Optimization (GRPO) in foundation LLMs, an increasing number of works have sought to adapt GRPO to Visual Large Language Models (VLLM…