Fetching the paper…

FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU · Around