BriefGPT.xyz
Mar, 2023
关于窃取语言模型解码算法的风险
On the Risks of Stealing the Decoding Algorithms of Language Models
HTML
PDF
Ali Naseh, Kalpesh Krishna, Mohit Iyyer, Amir Houmansadr
TL;DR
本研究发现黑客可以以低廉代价窃取现有文本API中流行的语言模型（包括GPT-2和GPT-3）的解码算法类型和超参数
Abstract
A key component of generating text from modern
language models
(LM) is the selection and tuning of
decoding algorithms
. These algorithms determine how to generate text from the internal probability distribution g
→