GPT-3: Language Models are Few-Shot Learners

Authors: Brown et al. Year: 2020 ArXiv/Link: https://arxiv.org/abs/2005.14165

Summary

A 175B parameter language model demonstrating few-shot learning capabilities across diverse tasks without task-specific fine-tuning.

Key Concepts

• Few-shot learning
• In-context learning
• Emergent abilities
• Task agnostic
• Large-scale pretraining

Impact

Demonstrated that scale alone enables powerful few-shot learning

Category

GPT Series

---