https://promptfoo.dev/ logo
Promptfoo With or Without Judge
# questions
j
Hi all, I have a question regarding to the use of Judge. i used redteam with some plugings (OWASP10...) and them i used redteam generate for generate prompts for test my LLM. I didnt use Judge and on the final report i observed some prompts that passed and other that failed. Whats the difference with Judge, the result go to be more accurate? Thanks in advance
Another question is the grader and the Judge the same?
.
i
grader and judge are the same in promptfoo 🙂
what do you mean by "didn't use Judge"?
j
Doesn´t matter i thought that should put "judge" and select a model for use as judge. Btw, how the graders are runing, by default i think that is a openAI model, some days ago i was testing Promptfoo using BurpSuite as Proxy for check what promptfoo send out of my network, and i didn´t see any prompt or answer for the model that i was testing sent to OpenAI