Unifying and Aligning Multifacted AI Explanations

Code Structure

• /src: Contains the TypeScript code used to generate the visualizations.
• /pyWidget contains the python code that create ML models, generate explanations, and call the visualization code.

Depedencies

• run npm install to install the dependencies for typescript

• conda env create -f environment.yaml to install the dependencies for python

How to Run

Dev Mode

1. Execute npm run widgetDev. The compiled TypeScript code will be stored in the /bundle directory. The hot module replacement will re-bundle the outputs when there is a change in src.
2. Activate the conda environment using conda activate mixai.
3. Launch the notebook pyWidget/explanation.ipynb. The Python code within will invoke the bundled TypeScript code to produce visualizations.

Product Mode

[place holder]