5 papers
Attention Sink Forges Native MoE in Attention Layers: Sink-Aware Training to Address Head Collapse
Zizhuo Fu, Wenxuan Zeng, Runsheng Wang +1
Large Language Models (LLMs) often assign disproportionate attention to the first token, a phenomenon known as the attention sink. Several recent approaches aim to address this iss…
UFO: Unlocking Ultra-Efficient Quantized Private Inference with Protocol and Algorithm Co-Optimization
Wenxuan Zeng, Chao Yang, Tianshi Xu +4
Private convolutional neural network (CNN) inference based on secure two-party computation (2PC) suffers from high communication and latency overhead, especially from convolution l…
H2EAL: Hybrid-Bonding Architecture with Hybrid Sparse Attention for Efficient Long-Context LLM Inference
Zizhuo Fu, Xiaotian Guo, Wenxuan Zeng +6
Large language models (LLMs) have demonstrated remarkable proficiency in a wide range of natural language processing applications. However, the high energy and latency overhead ind…
MPCache: MPC-Friendly KV Cache Eviction for Efficient Private LLM Inference
Wenxuan Zeng, Ye Dong, Jinjin Zhou +5
Private large language model (LLM) inference based on secure multi-party computation (MPC) achieves formal data privacy protection but suffers from significant latency overhead, es…
UniCAIM: A Unified CAM/CIM Architecture with Static-Dynamic KV Cache Pruning for Efficient Long-Context LLM Inference
Weikai Xu, Wenxuan Zeng, Qianqian Huang +2
Transformer-based large language models (LLMs) have achieved impressive performance in various natural language processing (NLP) applications. However, the high memory and computat…