6 papers
UniMoMo: Expert Merging-Based MoE Acceleration for Large Recommendation Models
Lei Xin, Bin Gu, Peize Li +9
Sparse mixture-of-experts (MoE) layers expand recommendation capacity through conditional computation, yet a trained checkpoint still stores and routes over its full expert bank. W…
MambaRaw: Selective State Space Modeling for Efficient 4K Raw Image Reconstruction
Peize Li, Fanhu Zeng, Tongda Xu +5
In-camera JPEG previews are ubiquitous in raw image formats and provide an sRGB reference at negligible storage cost. Although existing metadata-based reconstruction frameworks can…
Token Reduction Should Go Beyond Efficiency in Generative Models -- From Vision, Language to Multimodality
Zhenglun Kong, Yize Li, Fanhu Zeng +7
In Transformer architectures, tokens\textemdash discrete units derived from raw data\textemdash are formed by segmenting inputs into fixed-length chunks. Each token is then mapped…
MambaIC: State Space Models for High-Performance Learned Image Compression
Fanhu Zeng, Hao Tang, Yihua Shao +3
A high-performance image compression algorithm is crucial for real-time information transmission across numerous fields. Despite rapid progress in image compression, computational…
EventVAD: Training-Free Event-Aware Video Anomaly Detection
Yihua Shao, Haojin He, Sijie Li +11
Video Anomaly Detection~(VAD) focuses on identifying anomalies within videos. Supervised methods require an amount of in-domain training data and often struggle to generalize to un…
Token Transforming: A Unified and Training-Free Token Compression Framework for Vision Transformer Acceleration
Fanhu Zeng, Deli Yu, Zhenglun Kong +1
Vision transformers have been widely explored in various vision tasks. Due to heavy computational cost, much interest has aroused for compressing vision transformer dynamically in…