Revision History
Attention Is All You Need · 14 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 swapped dropout rates for the big model, the invented 41.0-vs-41.8 BLEU explanation and an overstated 10-100x compute claim, a fabricated Nature ranking comparison (COVID-19, AlphaFold) and unsupported superlative, the claim of a NeurIPS oral (it was a poster), a 2,000-attendee figure for the GTC 2024 panel (source: hundreds), Karpathy's 100x figure, and stale author roles (Vaswani now at NVIDIA, Shazeer to OpenAI, Parmar at Anthropic) (plus 31 minor fixes)
Version 14 · Nov 2, 2026, 08:28 AM
- Update: added a video.
Version 13 · Nov 1, 2026, 08:17 AM
- Improve internal links (2026-10-13)
Version 11 · Oct 13, 2026, 10:08 AM
- Added 6 contextual internal links
Version 10 · Jul 23, 2026, 03:04 PM
- Render formulas with LaTeX math notation
Version 9 · Jul 11, 2026, 06:09 PM