1 citations · 1 across the 6 of their papers we have counts for
7 papers
Faster Than Flash: Exploiting Attention Sparsity for Efficient Long-Context Decoding
Zhigeng Liu, Zhiyuan Ning, Ruixiao Li +5
The development of long-context Large Language Models (LLMs) is constrained by the memory bandwidth bottleneck and quadratic complexity of the attention mechanism during decoding.…
FeedbackTrack: Visual-Cortex-Inspired Cross-Frame Feedback for Transformer Tracking
Yueyang Cang, Xiaoteng Zhang, Zhiyuan Ning +2
Visual object tracking requires effective temporal integration, yet most Transformer trackers still rely on predominantly feed-forward feature extraction. Existing temporal mechani…
Recall Before You Rank: Similarity-Guided Top- Reuse for Efficient Long-Context Attention
Wenshuai Yao, Wenyong Zhou, Hanyong Shao +5
Top- sparse attention reduces the cost of Softmax and value aggregation by attending to only a small subset of key--value (KV) entries. However, identifying this subset still re…
MOVA: Towards Scalable and Synchronized Video-Audio Generation
OpenMOSS Team, Donghua Yu, Mingshu Chen +38
Audio is indispensable for real-world video, yet generation models have largely overlooked audio components. Current approaches to producing audio-visual content often rely on casc…
Breaking Model Lock-in: Cost-Efficient Zero-Shot LLM Routing via a Universal Latent Space
Cheng Yan, Wuyang Zhang, Zhiyuan Ning +5
The rapid proliferation of Large Language Models (LLMs) has led to a fragmented and inefficient ecosystem, a state of ``model lock-in'' where seamlessly integrating novel models re…
CAS-Spec: Cascade Adaptive Self-Speculative Decoding for On-the-Fly Lossless Inference Acceleration of LLMs
Zhiyuan Ning, Jiawei Shao, Ruge Xu +4
Speculative decoding has become a widely adopted as an effective technique for lossless inference acceleration when deploying large language models (LLMs). While on-the-fly self-sp…