Research breakthrough · curated entry
Attention Is All You Need proposes the Transformer
Attention Is All You Need
- Date
- June 12, 2017
- Category
- Research breakthrough
- Organization / Model
- Google Brain / Google Research
- Evidence
- 1 source · 1 evidence family
Vaswani and colleagues replaced recurrence entirely with self-attention, which made training highly parallel. The Transformer is the underlying architecture of GPT, BERT and essentially every large language model since.
Curator's note: Cross-checked against the arXiv paper.
Curated sources · 1
Cite this page
This page is a curated static snapshot and can be cited directly. Preferred format GB/T 7714—2025 (electronic resource [EB/OL]); the other 5 styles open below, together with the machine-readable files.
GB/T 7714—2025 (electronic resource [EB/OL])
AI 纪年. Transformer 架构提出，自注意力取代循环结构[EB/OL]. (2017-06-12)[2026-11-03]. https://aihist.net/e/transformer-2017.html.
APA 7 · American Psychological Association
AI 纪年. (2017, June 12). Transformer 架构提出，自注意力取代循环结构 [Web page]. https://aihist.net/e/transformer-2017.html
MLA 9 · Modern Language Association
"Transformer 架构提出，自注意力取代循环结构." AI 纪年, 12 June 2017, aihist.net/e/transformer-2017.html. Accessed 3 Nov. 2026.
Chicago 17 · notes-bibliography
AI 纪年. "Transformer 架构提出，自注意力取代循环结构." Published June 12, 2017. Accessed Nov. 3, 2026. https://aihist.net/e/transformer-2017.html.
IEEE · citation-reference list
"Transformer 架构提出，自注意力取代循环结构," AI 纪年, 12 June 2017. [Online]. Available: https://aihist.net/e/transformer-2017.html. [Accessed: 3 Nov. 2026].
ACM · author-date variant
AI 纪年. June 12, 2017. Transformer 架构提出，自注意力取代循环结构. Retrieved November 3, 2026 from https://aihist.net/e/transformer-2017.html.
All six styles come from one template and can be compared word for word. [引用日期], Accessed, urldate and Y2 all carry this page's build date, 2026-11-03 — a static page is an immutable snapshot, so substitute the date you actually consulted it. The address in every style is the archive's canonical page: this English page translates the same snapshot, and the work keeps its name, so one record cites both. For one-click copying use the same entry in the timeline, which takes the day you opened it. Expand the other styles before printing.