It seemed to be alluded to in the paper, but was wondering if the code for the automated benchmark generation will be open sourced. Thank you for your efforts!