Skip to content

Quick Start

1. Clone the Repository

git clone https://github.com/ophelia-rnd/bianor.git
cd bianor

2. Install Dependencies

pip install -r requirements.txt

3. Prepare the Demo Setup

The repository comes with a notebook that sets up local data and a representation model for the Iris dataset, along with a configured catalog.yaml manifest. Execute all cells in bianor_local_catalog/notebooks/demo_iris.ipynb.

4. Launch Bianor

Start the application and navigate to http://localhost:8050.

python app.py

5. Explore Bianor

  1. Select a Representation: Choose Iris Representation from the top navigation selector.
  2. Inspect Topology: Toggle between Neighbor Distances, Training Activations, and Topological Connections to inspect cluster structure and node density in 2D.
  3. Analyze Components: Use the Component View to visualize individual feature distributions across the map.
  4. Overlay Projections & Predictions: Display labeled historical samples and ground-truth classification of holdout observations.

Next Steps