Reducing Transformer Key-Value Cache Size with Cross-Layer Attention Paper • 2405.12981 • Published May 21, 2024 • 29 • 3