Evaluation from pandora files - selvaggi/mlpf GitHub Wiki

Start from the out_sim_edm4hep.root which is the output of the k4run CLDReconstruction.py ... command. Then use the mlpf/condor/make_pftree_clic_bindings.py script to convert the edm4hep data to a flat tree (our input format), like this:
python make_pftree_clic_bindings.py out_reco_edm4hep_REC.edm4hep.root tree.root True False. The third argument is what stores the pandora hits and info, if there are many events the script can take a very long time to run. If you use the data that was already generated with out pipelines this intermediate file is available in the same location as the edm4hep files from reconstruction.

Then to create the output file run this: python src/create_pandora_dataframe.py --data-test /eos/experiment/fcc/ee/datasets/mlpf/CLD/eval/120924/dr_05_10_15_v2/pf_tree_{1..2}.root --data-config config_files/config_hits_track_v4.yaml -clust -clust_dim 3 --num-workers 0 --gpus 3 --batch-size 10 --fetch-step 0.1 --condensation --tracks --correction --add-track-chis --predict --model-prefix /eos/user/m/mgarciam/datasets_mlpf/models_trained_CLD/test_pandora_output/

where model-prefix needs to be set to the directory where we want to store the output dataframe and --data-test is the path to the ML ready data from above.