Abstract page for arXiv paper 2608.27460: Accelerating LLM Inference via Vector Index Based Output Embeddings