Piotr Nawrot

Latest

• Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference