FastInferencePrototype

Fast Inference of Machine Learning Models following the ONNX standards in ROOT/TMVA SOFIE (“System for Optimized Fast Inference code Emit”)

Keras

Maps keras layers to equivalent graph operations and returns graph nodes connected with node inputs and corresponding nodes names and the model weights in list of numpy array format.

PyTorch

Converts a model scripted using Torchscript to its equivalent ONNX graph using dummy inputs, and utility functions. Using the ONNX graph, model data and weights are extracted.

tmva

Directory for Toolkit for Multivariate Analysis (TMVA) of the Root-Project. PyConverters: Functionality for converting PyTorch and keras models to .root files.

restructure

Restructured SOFIE to avoid dependency conflicts with required versions of Protobuf

serialisation

Serialising RModel and ROperators for saving them into ROOT format.

Collections

Collections of relevant documents, manuals, user-guides for the project.