...
- See all B-jet calibration meetings on Indico
- Framework experience (Prajita, July 6)
- Plans (Prajita, July 13)
- What needs to be added to JETM2 (August 17)
- Talk at Higgs workshop (October 4)
- Update from Brendon on, first training with large dataset and reasonable performance (October 23)
Planning
Samples and models on SDF
TDD output files:
Code Block |
---|
# ttbar non-all hadronic sample in PHYSVAL format; 600 GB
/gpfs/slac/atlas/fs1/d/pbhattar/BjetRegression/Input_Ftag_Ntuples/nonallhad_PHYSVAL_mc20_410470
# ttbar samples in FTAG format; 900 GB
/gpfs/slac/atlas/fs1/d/pbhattar/BjetRegression/Input_Ftag_Ntuples/Rel22_ttbar_DiLep
/gpfs/slac/atlas/fs1/d/pbhattar/BjetRegression/Input_Ftag_Ntuples/Rel22_ttbar_SingleLep
/gpfs/slac/atlas/fs1/d/pbhattar/BjetRegression/Input_Ftag_Ntuples/Rel22_ttbar_AllHadronic |
Umami-preprocessing datasets:
Code Block |
---|
# UPP_1: ttbar non-all had FTAG1, 1.5mil jets; 1 GB
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/FTag_upp/Rel22_ttbar_AllHadronic_bjetonly
# UPP_2: ttbar non-all had PHYSVAL, 100mil jets w/ 10 < pt_true; 133 GB
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/PHYSVAL_upp/Rel22_ttbar_NonAllHadronic/
# UPP_3: ttbar non-all had PHYSVAL, 49mil jets w/ 10 < pt_true < 50 GeV + 50mil jets w/ pt_true > 50 GeV; 112 GB
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/upp_output/physval/ttbar_allhad/half-half-100mil-bjet_resampled/ |
Trained models:
Code Block |
---|
### Trained on UPP_1
# No auxiliary tasks
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/models/GNX-bjr-ftag1_1p5mil_NoAux_21param_20230918-T082841
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/models/GNX-bjr-ftag1_1p5mil_NoAux_21param_20230918-T174839
### Trained on UPP_2
# No auxiliary tasks, trained in 2 separate batches
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/models/GNX-bjr-physval_80mil_NoAux_21param_v1_20231013-T121459
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/models/GNX-bjr-physval_80mil_NoAux_21param_v2_20231013-T121459
# With auxiliary tasks
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/models/GNX-bjr-physval_80mil_Aux_21param_20231023-T080935
### Trained on UPP_3
/gpfs/slac/atlas/fs1/d/bbullard/BjetRegression/models/GNX-bjr-physval_80mil_NoAux_21param_PtResampled |
SDF preliminaries
Compute resources
...