ExFold: Unified Expert Folding for Training-Free MoE Prefill-Decode Acceleration Paper • 2608.24938 • Published Aug 24 • 3
HYPIC: Accelerating Hybrid-Attention LLM Serving with Position-Independent Caching Paper • 2607.01299 • Published Jul 12 • 3
GDPevo: Evaluating Agent Self-Evolution on Real Business Tasks Paper • 2608.03764 • Published Aug 4 • 27