VLM

DIVA: Harnessing the Representation Divergence in Unified Multimodal Models for Mutual Reinforcement
TBD
Cite
WindowQuant: Mixed-Precision KV Cache Quantization based on Window-Level Similarity for VLMs Inference Optimization
Recently, video language models (VLMs) have been applied in various fields. However, the visual token sequence of the VLM is too long, …
Cite arXiv
WindowQuant: Mixed-Precision KV Cache Quantization based on Window-Level Similarity for VLMs Inference Optimization
Triage: Hierarchical Visual Budgeting for Efficient Video Reasoning in Vision-Language Models
Vision-Language Models (VLMs) face significant computational challenges in video processing due to massive data redundancy, which …
Cite arXiv IEEE
Triage: Hierarchical Visual Budgeting for Efficient Video Reasoning in Vision-Language Models