Exploiting Inter-Layer Expert Affinity for Accelerating Mixture-of-Experts Model Inference

Open in new window