Clarification on Evaluation Pipeline Usage for Jailbreaking Track

[Devansh Bhardwaj]

Dear Organizers,

I wanted to clarify whether it is permissible to use the provided evaluation pipeline as part of our approach in a feedback mechanism to enhance the prompt generation process. Would this be allowed within the rules of the competition?

Additionally, are we allowed to use OpenAI APIs for generation?

Thank you for your time and assistance.

Best regards,
Devansh

[Zhen Xiang]

Dear Participants,

Thank you for your question. We allow feedback mechanisms in the approach. However, the evaluation pipeline should be hardcoded in your implementation. Your results should be reproducible by running your submitted code without human interference. OpenAI API is acceptable if the cost is reasonable.

Best,

Organizers