Revision History
Cross-attention · 13 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected the description of Flamingo's cross-attention masking (each text token sees only the immediately preceding image), replaced an invented account of the DALL-E 2 decoder's conditioning with the paper's actual design, and qualified the claim that Llama and other popular LLMs contain no cross-attention layers (Llama 3.2-Vision uses a cross-attention adapter) (plus 19 minor fixes)
Version 13 · Nov 2, 2026, 04:10 AM
- Improve internal links (2026-10-13)
Version 11 · Oct 13, 2026, 10:09 AM
- Added 6 contextual internal links
Version 10 · Jul 23, 2026, 03:05 PM
- Render formulas with LaTeX math notation
Version 9 · Jul 11, 2026, 11:41 PM