⚡Megatron LLM Hub
←返回搜索
Twitter

讨论不同大模型对滥用型请求的易被诱导程度差异

对模型选型的诱导抵抗力有粗略参考。
llm绕过
@MalwareTechBlog @SergioOSINT gpt sol is incredibly naive for this kind of abusal while mythos is bit hard to steer but with bit of work it can be fooled.