Salt: Multimodal Multitask Machine Learning for High Energy Physics

High energy physics studies the fundamental particles and forces that constitute the universe, often through experiments conducted in large particle accelerators such as the Large Hadron Collider (LHC) (Evans & Bryant, 2008).Salt is a Python application developed for the high energy physics community that streamlines the training and deployment of advanced machine learning (ML) models, making them more accessible and promoting shared best practices.Salt features a generic multimodal, multitask model skeleton which, coupled with a strong emphasis on modularity, configurability, and ease of use, can be used to tackle a wide variety of high energy physics ML applications. Some key features ofSalt are listed below: Based on established frameworks: Salt is built upon PyTorch (Paszke et al., 2019) and Lightning (Falcon & The PyTorch Lightning team, 2019) for maximum performance and scalability with minimal boilerplate code. Multimodal, multitask models: Salt models support multimodal inputs and can be configured to perform various tasks such as classification, regression, segmentation, and edge classification tasks.Any combination of these can be used to flexibly define models for multitask learning problems. Customisable and extensible: Salt supports full customisation of training and model configuration through YAML config files.Its modular design allows for the easy integration of custom dataloaders, layers, and models. Train at scale: Salt can handle large volumes of data with efficient HDF5 (The HDF Group, 1997) dataloaders.It also includes multi-GPU support from Lightning, enabling distributed training. Deployment ready: Salt facilitates ONNX (Bai et al., 2019) serialisation for integrating models into C++ based software environments.