Small preprocessing changes

umami/configs/GNN_Variables.yaml

@@ -35,6 +35,7 @@ train_variables:
     - numberOfSCTSharedHits
     - numberOfPixelHoles
     - numberOfSCTHoles
+    #- leptonID
     #- AMVFWeightPV
     #- ambiRank
     #- chiSquaredOverNumberDoF
@@ -44,5 +45,8 @@ track_train_variables:
   tracks:
     <<: *tracks_variables
 
+  tracks_loose:
+    <<: *tracks_variables
+
 
 custom_defaults_vars:

umami/data_tools/Cuts.py

@@ -47,6 +47,9 @@ def GetSampleCuts(jets, cuts):
     cuts = clean_list
 
     for cut_entry in cuts:
+        if cut_entry is None:
+            continue
+
         # expect a dictionary with only one entry
         cut = list(cut_entry.keys())
         if len(cut) != 1:

umami/preprocessing_tools/Preparation.py

@@ -90,7 +90,7 @@ class PrepareSamples:
             logger.warning("no batch size given. Batch size set to 1,000,000")
             self.batch_size = 1_000_000
         else:
-            self.batch_size = self.config.preparation["batchsize"]
+            self.batch_size = int(self.config.preparation["batchsize"])
 
     def GetBatchesPerFile(self, filename: str):
         """

umami/preprocessing_tools/Resampling.py

@@ -340,6 +340,9 @@ class Resampling:
         self.outfile_name = self.config.GetFileName(option="resampled")
         self.outfile_path = self.config.config["parameters"]["sample_path"]
 
+        if os.path.dirname(self.outfile_name):
+            os.makedirs(os.path.dirname(self.outfile_name), exist_ok=True)
+
         # Get class labels from sampling/preparation.
         # Try/Except here for backward compatibility
         try:
@@ -601,7 +604,6 @@ class Resampling:
 
             if create_file:
                 create_file = False
-                os.makedirs(os.path.dirname(self.outfile_name), exist_ok=True)
                 # write to file by creating dataset
                 with h5py.File(self.outfile_name, "w") as out_file:
                     out_file.create_dataset(