Skip to content
Longterm Wiki

Scaling Laws

Capabilities Researchactive

Empirical research on how AI capabilities scale with compute, data, and parameters, crucial for forecasting future capabilities.

Organizations
3
Key Papers
2
Grants
2
Total Funding
$89K
First Proposed: 2020 (Kaplan et al., OpenAI)
Cluster: Capabilities Research

Tags

scalingcapabilitiesforecasting

Organizations3

OrganizationRole
Anthropicactive
Google DeepMindactive
OpenAIpioneer

Grants2

NameRecipientAmountFunderDate
4-month stipend: Research on agent scaling laws—relationships between training compute and agent capabilities of LLMsAxel Højmark$70KLong-Term Future Fund (LTFF)2024-07
10-month salary for research on AI safety/alignment, scaling laws, and potentially interpretabilityBenedikt Hoeltgen$19KLong-Term Future Fund (LTFF)2021-10

Funding by Funder

FunderGrantsTotal Amount
Long-Term Future Fund (LTFF)2$89K

Key Papers & Resources2

SEMINAL
Scaling Laws for Neural Language Models
Kaplan et al. (OpenAI)2020
SEMINAL