Google Research unveiled TurboQuant, a novel quantization algorithm that compresses large language models’ Key-Value caches ...