responsible-ai-toolbox
responsible-ai-toolbox copied to clipboard
microsoft
Responsible AI Toolbox is a suite of tools providing model and data exploration and assessment user interfaces and libraries that enable a better understanding of AI systems. These interfaces and libr...
Running the latest version on Jupyterhub as an LDAP logged in user, the ErrorAnalysisDashboard gives a black screen. Attempting to open it in a separate tab also doesn't show anything:...
The MetricAPI can support custom metrics - would it be possible to add a user specific metric as a passable option into the dashboard? For example, rather than add Word...
Need by_group (currently in MetricFrame) in the Fairness dashboard. Found a temporary workaround (described below) ``` #sensitive_features is a list of the sensitive columns in the pandas dataframe (named pdf),...
It would be great if one can select multiple cohorts from the global explanation to show in the lower dependence plot: 
Currently, it is not possible to compare the distribution of different cohorts. We could possible reuse the design of global explanations to allow for comparative views: 
When the data is large and we are downsampling for explanations: - The data explorer should work for the whole data but global and local explanations only for the sampled...
Not sure what the priority of this should be. Currently if you do some analysis and get to a particular view within the Error Analysis widget. And then click on...
Interpret dashboard has two top-level components, ExpanationDashboard and NewExplanationDashboard. When the new dashboard is determined to be sufficient, the old dashboard and its components should be moved to a legacy...
## Description Unifying the true_y and predicted_y tests. ## Checklist - [ ] I have added screenshots above for all UI changes. - [ ] I have added e2e tests...
## Description Seems like we weren't passing the classes from rai_insights to cohort filtering class. ## Checklist - [ ] I have added screenshots above for all UI changes. -...