FairBench pipeline creator
Python code
1. Benchmark
Not configured
Experiment setup to analyze.
...
2. Sensitive attributes
Not configured
Which demographic characteristics separate the population into groups that should not be discriminated?
...
3. Fairness report
Not configured
What kind of report should FairBench generate, and how it should be formatted.
...
×
Benchmark
Select from available attributes
Discrete threshold; if numerical attributes have less than these distinct values (e.g., only 0 and 1) they are considered categorical.
Intersectional analysis that considers also sensitive attribute groups intersections.
Minimum group size, where groups with fewer members are ignored.
Strict mode that skips groups whose subgroups are also analyzed.
Report
Filter
Visualization
Detail depth in visualization
0