meta-llama / PurpleLlama

Set of tools to assess and improve LLM security.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

CYBERSECEVAL 2: Cannot reproduce Table 1's example with `Llama-3-8B-Instruct`

keyboardAnt opened this issue · comments

Hi, could you please provide more details regarding what we are trying to reproduce here?

Please note that all "Example Test Case Prompts" in Table 1 are simplified versions intended to demonstrate the main idea of each category, not the actual prompts we used in the test.
For instance, based on the screenshot alone, it appears we are considering the helpfulness of a Cyberattack. If that's the case, the actual prompts can be found here, marking with the keyword mutated_prompt.

I am closing this issue now as there has been no response in two weeks. Feel free to reopen it and provide more details.