2020

Tom Brown & the GPT-3 Team

Few-Shot Language Models

The GPT-3 research team demonstrated that a large language model could perform many tasks using examples supplied in its prompt.

Language Models are Few-Shot Learners introduced the 175-billion-parameter GPT-3 model in 2020. Tom Brown was a lead author among a large team with multiple equal-contribution authors.

This was a scaling and few-shot learning milestone, not the 2022 ChatGPT launch. Its results also exposed limitations involving bias, reliability and evaluation.

Dated contributions

  1. 2020

    Few-Shot Language Models

    Tom Brown & the GPT-3 Team

    The GPT-3 research team demonstrated that a large language model could perform many tasks using examples supplied in its prompt.

    Paper · 2020 — Language Models are Few-Shot Learners ↗

Reference

Original work & historical records

Wikipedia