Laboratory-Scale AI: Open-Weight Models are Competitive with ChatGPT Even in Low-Resource Settings
Jun 1, 2024·,,
,,,,,,,·
1 min read
Robert Wolfe
Isaac Slaughter
Bin Han
Bingbing Wen
Yiwei Yang
Lucas Rosenblatt
Bernease Herman
Eva Maxfield Brown
Zening Qu
Nic Weber
Bill Howe
Abstract
We show that carefully selected open-weight language models can achieve performance competitive with proprietary systems like ChatGPT even in low-resource institutional settings, enabling more equitable access to AI.
Type
Publication
FAccT 2024
We study how open-weight models can be deployed in low-resource settings, finding that with appropriate selection and configuration they can rival proprietary systems such as ChatGPT for many practical tasks.
