Question regarding the number of runs

[ybendou]

Hi,

Thank you for the very nice repo! I have a question regarding the number of runs you use to generate your results as it is not very clear in the paper. How many few-shot evaluation runs are used per datapoint?

Thanks in advance!