Organisation

This guide will help you navigate through the various folders and components within it.

Main Repository Folder: etx4velo

The primary folder of this repository, etx4velo, serves as the hub for models, notebooks, and scripts related to our project.

Here’s a brief overview of what you’ll find:

• pipeline: This Python package is the heart of our project. It defines all the PyTorch models and includes utilities for data processing and plotting.

• pipeline_configs: In this folder, you’ll discover YAML files. Each YAML file corresponds to a comprehensive configuration for one training and inference pipeline.

• notebooks: This folder contains Jupyter notebooks, to train and test the full pipeline in an interactive manner.

• scripts: If you prefer script-based execution, this folder offers scripts for every step of the pipeline, from data preprocessing to track finding evaluations.

• artifact_archive: This folder is an archive of important runs: model checkpoints, reports and configurations.

• snakefiles: In this directory, you’ll discover Snakemake files. These files include rules designed to automatically generate plots used in presentations and conferences, ensuring reproducibility.

Additional repository Folders

Beyond the etx4velo folder, there are a few other directories:

• docs: this folder holds documentation related to our project.

• readme: In this folder, you’ll find essential setup guides to get you started

• montetracko: this folder is a Git submodule containing the montetracko library.

• setup: here, you’ll find various files crucial for setting up the repository.

Using Notebooks and Snakemake

The Jupyter notebooks, located in etx4velo/notebooks, are essential for training and testing. Additionally, we employ the The Snakemake workflow management system to generate figures in a reproducible manner from trained models.