11 papers
CoSA: Accelerating Long-Context Inference via Proxy-Kernel Co-Designed Sparse Attention
Yufei Xue, Lin Niu, Hong Liu +6
The quadratic cost of self-attention makes long-context inference prohibitively expensive, and proxy-based block-sparse attention has become a practical remedy. Existing methods ty…
MeanFlowNFT: Bringing Forward-Process RL to Average-Velocity Generators
Yushi Huang, Xiangxin Zhou, Jun Zhang +2
MeanFlow generators achieve fast few-step sampling by predicting average velocities over time intervals, making them attractive for efficient generation. Reinforcement learning (RL…
LENS: A Staged Design for Interaction Granularity in Sequential CTR Prediction
Yuan Wang, Yue Liu, Jun Zhang +1
In sequential CTR prediction, a central design question is at what granularity the target should interact with the user behaviour sequence. Existing models mainly follow two routes…
Reinforcing Few-step Generators via Reward-Tilted Distribution Matching
Yushi Huang, Xiangxin Zhou, Ruoyu Wang +3
Recent advances in few-step diffusion distillation have enabled efficient image generation, yet aligning these models with human preferences remains challenging. We propose Reward-…
The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence
Aili Chen, Aonian Li, Baichuan Zhou +215
We introduce the MiniMax-M2 series, a family of Mixture-of-Experts language models built around the principle that mini activations can unleash maximum real-world intelligence. The…
Salt: Self-Consistent Distribution Matching with Cache-Aware Training for Fast Video Generation
Xingtong Ge, Yi Zhang, Yushi Huang +6
Distilling video generation models to extremely low inference budgets (e.g., 2--4 NFEs) is crucial for real-time deployment, yet remains challenging. Trajectory-style consistency d…
Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing
Lingkun Long, Yushi Huang, Shihao Bai +4
Diffusion Large Language Models (dLLMs) deliver strong long-context processing capability in a non-autoregressive decoding paradigm. However, the considerable computational cost of…
MoDES: Accelerating Mixture-of-Experts Multimodal Large Language Models via Dynamic Expert Skipping
Yushi Huang, Zining Wang, Zhihang Yuan +5
Mixture-of-Experts (MoE) Multimodal large language models (MLLMs) excel at vision-language tasks, but they suffer from high computational inefficiency. To reduce inference overhead…
LinVideo: A Post-Training Framework towards O(n) Attention in Efficient Video Generation
Yushi Huang, Xingtong Ge, Ruihao Gong +2
Video diffusion models (DMs) have enabled high-quality video synthesis. However, their computation costs scale quadratically with sequence length because self-attention has quadrat…
SalQ-VLM: Fine-Grained Saliency-Guided Quantization for Vision-Language Models
Yufei Xue, Yushi Huang, Jiawei Shao +4
Large language models (LLMs) have demonstrated remarkable capabilities across diverse language tasks, motivating their extension to vision-language models (VLMs) for multimodal und…
QVGen: Pushing the Limit of Quantized Video Generative Models
Yushi Huang, Ruihao Gong, Jing Liu +4
Video diffusion models (DMs) have enabled high-quality video synthesis. Yet, their substantial computational and memory demands pose serious challenges to real-world deployment, ev…