EventClassifierTransformer

Library for Event Classifier Transformer using pytorch. See [https://arxiv.org/abs/2401.00428] for more information. Each library file (except Disco.py) can be ran individually, which executes an example of using the library.

Files

• EventClassifierTransformer.py: Implementation of Event Classifier Transformer.
• LossFunction.py: Library for loss functions. Contains extreme disco loss and bce disco loss.
• EvaluateModel.py: Selects best trained model based on binned significance. Makes an folder called output.
• RootDataset.py: Library for handling ROOT files inside pytorch. Inspired from [https://github.com/jmduarte/capstone-particle-physics-domain]
• Disco.py: Disco loss function from [https://github.com/gkasieczka/DisCo]

Example data for code

• data/example_data.root

Example model for code

• model_weights/EventClassifierTransformer_weights.pt