5 papers
MXAttention: Data-Free Optimal Scaling and Pre-Normalization Quantization for MXFP4 Attention
Jianlin Yu, Jing Lin, Linghui Kong +13
The quadratic cost of attention is a major bottleneck in diffusion-based video generation models. MXFP4 attention provides a promising path toward efficient inference, but direct M…
RainFusion2.0: Temporal-Spatial Awareness and Hardware-Efficient Block-wise Sparse Attention
Aiyue Chen, Yaofu Liu, Junjian Huang +6
In video and image generation tasks, Diffusion Transformer (DiT) models incur extremely high computational costs due to attention mechanisms, which limits their practical applicati…
MEPIC: Memory Efficient Position Independent Caching for LLM Serving
Qian Wang, Zahra Yousefijamarani, Morgan Lindsay Heisler +8
Modern LLM applications such as deep-research assistants, coding agents, and Retrieval-Augmented Generation (RAG) systems, repeatedly process long prompt histories containing share…
AnyExperts: On-Demand Expert Allocation for Multimodal Language Models with Mixture of Expert
Yuting Gao, Wang Lan, Hengyuan Zhao +3
Multimodal Mixture-of-Experts (MoE) models offer a promising path toward scalable and efficient large vision-language systems. However, existing approaches rely on rigid routing st…
EvoMoE: Expert Evolution in Mixture of Experts for Multimodal Large Language Models
Linglin Jing, Yuting Gao, Zhigang Wang +5
Recent advancements have shown that the Mixture of Experts (MoE) approach significantly enhances the capacity of large language models (LLMs) and improves performance on downstream…