Metrics¶
Metrics are the way to numerically measure some desired or undesired properties of an object. Common example is performance metrics in machine learning. They are defined to show quality of the solution. When computed on various validation sets they can provide engineers with information on what to expect from the model in real-life settings.
In XAIB each metric should correspond to some Case
To further read on metrics in this benchmark, please proceed to the following pages.
To see how different XAI methods perform on the different datasets, please proceed to the results page.