6 papers
RAC: Reference-Aware Activation Compression for Communication-Efficient Split LLM Inference
Guotao Yang, Mingxi Zhao, Haopeng Li +4
Large language model (LLM) agents repeatedly process long, privacy-sensitive contexts, while cloud-only deployment exposes user data beyond the trusted endpoint and fully local dep…
AsymSpec: Efficient Cloud-Edge Speculative Decoding over Asymmetric Networks
Guotao Yang, Hao Chen, Rui Guo +5
Cloud-edge speculative decoding places a lightweight draft model at an edge gateway and a higher-quality target model in the cloud, but inserts communication into every speculative…
PAT: Accelerating LLM Decoding via Prefix-Aware Attention with Resource Efficient Multi-Tile Kernel
Jinjun Yi, Zhixin Zhao, Yitao Hu +7
LLM serving is increasingly dominated by decode attention, which is a memory-bound operation due to massive KV cache loading from global memory. Meanwhile, real-world workloads exh…
Mosaic: Unlocking Long-Context Inference for Diffusion LLMs via Global Memory Planning and Dynamic Peak Taming
Liang Zheng, Bowen Shi, Yitao Hu +5
Diffusion-based large language models (dLLMs) have emerged as a promising paradigm, utilizing simultaneous denoising to enable global planning and iterative refinement. While these…
RAGPulse: An Open-Source RAG Workload Trace to Optimize RAG Serving Systems
Zhengchao Wang, Yitao Hu, Jianing Ye +4
Retrieval-Augmented Generation (RAG) is a critical paradigm for building reliable, knowledge-intensive Large Language Model (LLM) applications. However, the multi-stage pipeline (r…
Harpagon: Minimizing DNN Serving Cost via Efficient Dispatching, Scheduling and Splitting
Zhixin Zhao, Yitao Hu, Ziqi Gong +5
Advances in deep neural networks (DNNs) have significantly contributed to the development of real-time video processing applications. Efficient scheduling of DNN workloads in cloud…