Just dropping an FYI for the authors/slash any innocent soul that is potentially attempting to apply this methodology to new data. Not only is it impossible to replicate the authors' results as instructed in the README, it is even more impossible to generate results on new data, as the pipeline depends on files that ought to be created but are not, among numerous other bugs that I won't even attempt to document here. This seems like a very cool paper, so it's a bummer that it's currently in the state it is. Hope it gets renovated at some point in the future!
Just dropping an FYI for the authors/slash any innocent soul that is potentially attempting to apply this methodology to new data. Not only is it impossible to replicate the authors' results as instructed in the
README
, it is even more impossible to generate results on new data, as the pipeline depends on files that ought to be created but are not, among numerous other bugs that I won't even attempt to document here. This seems like a very cool paper, so it's a bummer that it's currently in the state it is. Hope it gets renovated at some point in the future!