Running a quick example

This page shows how to get RAITAP running quickly, with a simple pre-defined example. If you want to see how to fully configure your own assessment, skip to the Creating & running your own configuration page.

Note

This page assumes you have already installed RAITAP. If you didn’t, see the Installation page.

1. Install dependencies

Our pre-defined examples uses a PyTorch model and runs a transparency assessment using Captum. Hence, we install the dependencies:

uv

uv add "raitap[captum,reporting,torch-cpu]"

pip

pip install "raitap[captum,reporting,torch-cpu]"

Note

The example is light enough to run on a CPU, hence we used torch-cpu. Feel free to use another execution profile (e.g. torch-cuda). Refer to Execution dependencies for more details.

2. Run the example

Our pre-defined example is the default config shipped with RAITAP. This means you do not need to specify any options to run it.

uv

uv run raitap

pip

raitap

3. Inspect the output

After the run is complete, the outputs directory can be found in the directory you ran RAITAP from.

It will contain the run’s metadata, as well as the transparency assessment (attributions and visualisations). Refer to the Understanding outputs page for more details.