{"type":"link","version":"1.0","title":"Sparse attention keeps full attention's expressive power in theory but some tasks then need many more layers","author_name":"AI Archs","author_url":"https://ai-arch.pages.dev","provider_name":"AI Archs","provider_url":"https://ai-arch.pages.dev","url":"https://ai-arch.pages.dev/n/sparse-attention-preserves-expressivity-but-can-demand-more-layers","thumbnail_url":"https://ai-arch.pages.dev/og/sparse-attention-preserves-expressivity-but-can-demand-more-layers.png","thumbnail_width":1200,"thumbnail_height":630}