Skip to content

Changing the PbPb NN weights for ghost probability calculation to the newest version

Ivan Cambon Bouzas requested to merge jcambonb_PbPb_gprob_v2 into 2024-patches

Due to the problems seeing at the pit, some fix to the NN weights for PbPb were required to have a proper good distribution of track ghost probability. For that, new weights were trained with some fix to the flattening problem. This new weights are labeled as v2. The changes made at Moore level was only updating the ReconConf/hlt2_tracking.py script

track_ghost_prob_distributions_PbPb_weights_v2.pdf

track_ghost_prob_distributions_nout_PbPb_weights_v2.pdf

This MR is linked to lhcb-datapkg/ParamFiles!115 (merged)

Edited by Ivan Cambon Bouzas

Merge request reports