Skip to content

Fixes for dedicated sample study and small event size case

Jay Chan requested to merge jay_small_fix into dev

This MR introduces fixes to:

  • Account for the extreme case where event size is very small (e.g. single-particle).
  • Handle cases where certain dataset (e.g. trainset and valet) is set to 0 (i.e. we are only interested in inference)
  • Use the correct way to calculate the number of shared edges

Also add features:

  • Option to configure train/val/test split used for athena_reader in the data_reader yaml file

I have tested it with example 1 and single particle events. Both work fine

Edited by Jay Chan

Merge request reports