Skip to main content
PaperLLMsv1.0

Training Compute-Optimal Large Language Models (Chinchilla)

by DeepMind · free · Last verified 2026-03-17

Challenges the Kaplan et al. scaling laws by showing that model size and training tokens should scale equally. Trains Chinchilla (70B) on 4× more data than Gopher, matching or beating models 4× its size, redefining compute-optimal training strategies.

https://arxiv.org/abs/2203.15556
B+
B+Good
Adoption: AQuality: A+Freshness: BCitations: AEngagement: F

Specifications

License
Open Access
Pricing
free
Capabilities
language-modeling, reasoning, scaling-analysis
Integrations
Use Cases
language-modeling, compute-optimal-training, research
API Available
No
Tags
chinchilla, scaling-laws, compute-optimal, deepmind, training, foundational
Added
2026-03-17
Completeness
100%

Index Score

75.4
Adoption
85
Quality
97
Freshness
62
Citations
88
Engagement
0

Put AI to work for your business

Deploy this paper alongside autonomous AaaS agents that handle tasks end-to-end — no babysitting required.

Explore the full AI ecosystem on Agents as a Service