In-Context Autoencoder for Context Compression in a Large Language Model figure
AlphaXiv 中文概览(可滚动查看)