You don't have to believe. The repository is there for anyone to attempt reproducing the results. Criticisms without proof when there's a pretty straightforward way toward that proof are pointless. Go run the experiment and rip that apart if it doesn't hold up. And until then, refrain from criticizing.