HHbtag training

HHbtag is the HH-optimised b-jet identification used by the analysis. This page covers the specialised workflow of producing the training skim — only needed if you are (re)training the tagger, not for normal analysis runs.

Produce the training skim

python Studies/HHBTag/CreateTrainingSkim.py \
  --inFile  <NANOAOD_INPUT>.root \
  --outFile output/skim.root \
  --mass 350 --sample GluGluToBulkGraviton --year 2018 >& EventInfo.txt

Adapt the example to your inputs

The arguments above (--mass, --sample, --year, the input path) are an illustration. Point --inFile at the NanoAOD signal sample you want to train on and set --mass/--sample/--year accordingly. Check CreateTrainingSkim.py --help for the current options.

The skim is the input to the HHbtag training itself, which lives in the HHbtag repository.

You usually don't need this

Standard analysis productions use the already-trained HHbtag model that ships with the HHbtag submodule. Retraining is an expert task done rarely.