long prompt evaluation

[ywlq]

I am using the augment's long prompt for evaluation. Since the prompt is too long, I'm using the corresponding short prompt as the filename. When conducting the evaluation, should I directly use the default prompt_file from evaluate_long.sh, or do I need to customize the prompt_file? If customization is needed, what should be the format of the prompt_file?

[ziqihuangg]

Hi, for the filename, please use the original short prompt directly, since that’s what the evaluation pipeline will look up.