HGCalML

Requirements

• DeepJetCore 3.X (https://github.com/DL4Jets/DeepJetCore)
• DeepJetCore 3.X container (or latest version in general)

For CERN (or any machine with cvmfs mounted), a script to start the latest container use this script:

#!/bin/bash

gpuopt=""
files=$(ls -l /dev/nvidia* 2> /dev/null | egrep -c '\n')
if [[ "$files" != "0" ]]
then
gpuopt="--nv"
fi

#this is a singularity problem only fixed recently
unset LD_LIBRARY_PATH
unset PYTHONPATH
sing=`which singularity`
unset PATH
cd

$sing run -B /eos -B /afs $gpuopt /cvmfs/unpacked.cern.ch/registry.hub.docker.com/cernml4reco/deepjetcore3:latest

The package follows the structure and logic of all DeepJetCore subpackages (also the example in DeepJetCore). So as a fresh starting point, it can be a good idea to follow the DeepJetCore example first.

Setup

git clone  --recurse-submodules  https://github.com/cms-pepr/HGCalML
cd HGCalML
source env.sh #every time
./setup.sh #just once, compiles custom kernels

When developing custom CUDA kernels

The kernels are located in modules/compiled The naming scheme should be obvious and must be followed. Compile with make.

Converting the data from ntuples

convertFromSource.py -i <text file listing all training input files> -o <output dir> -c TrainData_NanoML The conversion rule itself is located here: modules/datastructures/TrainData_NanoML.py

The training files (see next section) usually also contain a comment in the beginning pointing to the latest data set at CERN and flatiron.

Standard training and inference

Go to the Train folder and then use the following command to start training. The file has code for running plots and more. That can be adapted according to needs.

cd Train

Look at the first lines of the file std_training.py containing a short description and where to find the dataset compatible with that training file. Then execute the following command to run a training.

python3 std_training.py <path_to_dataset>/training_data.djcdc <training_output_path>

Please notice that the standard configuration might or might not include writing the printout to a file in the training output directory.

For inference, the trained model can be applied to a different test dataset. Please note that this is slightly different from the standard DeepJetCore procedure.

predict_hgcal.py <training_output_path>/KERAS_model.h5  <path_to_dataset>/testing_data.djcdc  <inference_output_folder>

To analyse the prediction, use the analyse_hgcal_predictions.py script.