# Language Models are Few-Shot Learners

> GPT-3 demonstrates that scaling language models to 175 billion parameters enables strong few-shot performance on diverse tasks without gradient updates.

- **Source:** [arXiv](https://arxiv.org/abs/2005.14165)
- **Published:** 2026-03-07
- **Permalink:** https://picx.dev/p/1pZt28
- **Whiteboard:** https://picx.dev/p/1pZt28/image

## Summary

# Summary of "Language Models are Few-Shot Learners"

## Summary (Overview)
*   **Scaling to 175B Pa

---

_Markdown view of https://picx.dev/p/1pZt28, served by PicX — AI-generated visual whiteboard summaries of research papers._