thu-coai / Safety-Prompts

Chinese safety prompts for evaluating and improving the safety of LLMs. 中文安全prompts,用于评估和提升大模型的安全性。
http://coai.cs.tsinghua.edu.cn/leaderboard/
Apache License 2.0
872 stars 81 forks source link

What model is the LLM used in Figure 3? #20

Closed XiaoluJiayou closed 8 months ago

XiaoluJiayou commented 8 months ago

hello, What model is the LLM used in Figure 3 of the paper to determine whether the conversation is safe, and does it require retraining or fine-tuning? image

TissueC commented 8 months ago

在当时使用的是InstructGPT,但放到今天来看这个选择显然不是最优了。