This repository contains the code required to reproduce the current paper plots.
All scripts rely on .csv-files containing retro, gnn and truth variables. These are available on cobalt at: /data/user/rorsoe/paper_data
Note: Computing classification and performance plots is quick. Computing robustness plots directly from the raw .csv-files takes a long time. Therefore, a summary df containing the derived statistics is included here together with the code that produceses it. You may alter it as you wish.
Rasmus