AI ChronicleAI 纪年 Full timeline →

Model release · curated entry

GPT-3 makes few-shot prompting a general capability

Language Models are Few-Shot Learners

Date
May 28, 2020
Category
Model release
Organization / Model
OpenAI
Evidence
2 sources · 2 evidence families

OpenAI released the 175-billion-parameter GPT-3, showing that in-context few-shot learning generalises across tasks. Prompting became the main way people interact with large models.

Curator's note: Cross-checked against the arXiv paper and OpenAI's official page.

Curated sources · 2

  1. Paper preprints · arXiv2020-05-28

    Language Models are Few-Shot Learners

    arxiv.org

  2. Official announcement · OpenAI2020-05-28

    Language Models are Few-Shot Learners

    openai.com

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. GPT-3 发布,少样本上下文学习成为通用能力[EB/OL]. (2020-05-28)[2026-11-03]. https://aihist.net/e/gpt-3-2020.html.

APA 7 · American Psychological Association

AI 纪年. (2020, May 28). GPT-3 发布,少样本上下文学习成为通用能力 [Web page]. https://aihist.net/e/gpt-3-2020.html

MLA 9 · Modern Language Association

"GPT-3 发布,少样本上下文学习成为通用能力." AI 纪年, 28 May 2020, aihist.net/e/gpt-3-2020.html. Accessed 3 Nov. 2026.

Chicago 17 · notes-bibliography

AI 纪年. "GPT-3 发布,少样本上下文学习成为通用能力." Published May 28, 2020. Accessed Nov. 3, 2026. https://aihist.net/e/gpt-3-2020.html.

IEEE · citation-reference list

"GPT-3 发布,少样本上下文学习成为通用能力," AI 纪年, 28 May 2020. [Online]. Available: https://aihist.net/e/gpt-3-2020.html. [Accessed: 3 Nov. 2026].

ACM · author-date variant

AI 纪年. May 28, 2020. GPT-3 发布,少样本上下文学习成为通用能力. Retrieved November 3, 2026 from https://aihist.net/e/gpt-3-2020.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.