KV Cache and Flash Attention with interactive diagrams

An interactive website visually explains the concepts of KV Cache and Flash Attention, two key optimization techniques used in transformer-based language models to improve inference efficiency and memory usage.