hyperparameter for pass@k inference

Hi authors, I've been wanting to reproduce the performance, but the paper does not reveal anything about the hyperparameters used in the experiments. Could you provide more information on the inference setting? thanks