// ▲ Pretraining, Scaling & Distributed Training

Emergent Abilities

Some abilities appear abruptly only past a certain scale.

涌现能力 2022 Foundational core

IN PLAIN WORDS

Tasks like multi-step arithmetic stay near random for small models then jump at scale. Later work argued this is partly an artifact of discontinuous metrics, so emergence remains debated.

Used in: Capability forecasting debates

TIMELINE · prerequisites → this → what builds on it

2020 Scaling Laws2022 Emergent Abilities

01Original paper

Emergent Abilities of Large Language Models

Jason Wei, Yi Tay, Rishi Bommasani, et al. · 2022

arXiv ↗ ✓ Verified · arXiv API

02Where it sits

Learn the left column first; the right column builds on this one.

Learn first

Scaling Laws

Emergent Abilities

Then you can learn

Nothing in the atlas builds on it yet

Related:Chain-of-Thought Prompting

Left: prerequisites · right: builds on this · dashed: related. Full map →

03Learning resources

Official docs, explainers, courses and reference code. All links checked.

Courses

cs336.stanford.edu ↗EN

04Where the course teaches it

U44 Pretraining, Data & Scaling Laws Pre-training data pipelines, scaling laws, Chinchilla, μP, MTP and training stability. 10 h6 algorithms4 papers U90 Evaluation & Benchmarks Major benchmarks, arenas, data contamination and the limits of capability evaluation. 6 h2 algorithms2 paperslesson ready

For AI / data

Verification: verified, checked 2026-10-21.
JSON: /api/learn/algorithms.json (id: emergence)