Prefix-Aware Attention for LLM Decoding
Python 54 3
An Open-Source RAG Workload Trace to Optimize RAG Serving Systems
Python 45 4
C++ 81 110
[ICML'26] MOSAIC: Unlocking Over 30× Context Length for Diffusion LLMs Inference via Global Memory Planning and Dynamic Peak Taming
Faster Gradient Computation for Delta-Rule Attention Family
Accelerating MoE-Based dLLM Inference via Inter-Iteration Locality-Aware Expert Offloading