I didn't see the specific task performance of BBH or the optimized prompts in the paper.
Could you provide the final prompts for each BBH task or the optimization scripts used on the BBH dataset?
The "BBH" folder in the provided code should be correctly renamed to "II", as it does not appear to use the BBH and BBII dataset.