training.py

 
 
import os
import os.path
import shutil
import subprocess
 
from tracking.run.mixins import BrowseTFileOnTerminateRunMixin
from tracking.run.mixins import PostProcessingRunMixin
from tracking.run.utilities import NonstrictChoices
 
 
class TrainingRunMixin(BrowseTFileOnTerminateRunMixin, PostProcessingRunMixin):
    """Prepare and execute a basf2 job to train neural network, postprocess, and inspect"""
 
    
    task = "train"
 
    
    variables = None
 
    
    groupby = None
 
    
    auxiliaries = None
 
    
    truth = "truth"
 
    @property
    def identifier(self):
        """Database identifier of the filte being trained"""
        return "trackfindingcdc_" + self.__class__.__name__[:-len("TrainingRun")]
 
    @property
    def sample_file_name(self):
        """File name of the recorded sample to be trained on
 
        Defaults to the class name minus the mandatory TrainingRun postfix
        """
        return self.__class__.__name__[:-len("TrainingRun")] + '_' + self.tasktasktask + '.root'
 
    def create_argument_parser(self, **kwds):
        """Create argument parser"""
        argument_parser = super().create_argument_parser(**kwds)
 
        argument_parser.add_argument(
            "--task",
            choices=NonstrictChoices(["train", "eval", "explore", ]),
            default=self.tasktasktask,
            dest="task",
            help=("Select a prepared recording task")
        )
 
        return argument_parser
 
    def postprocess(self):
        """Run the training as post-processing job
 
        To run only the training run with --postprocess-only
        """
        
        if self.tasktasktask == "train":
            cmd = [
                "trackfindingcdc_teacher",
            ]
 
            if self.variablesvariables:
                cmd += ["--variables"]
                cmd += self.variablesvariables
 
            cmd += [
                "--identifier=" + self.identifieridentifier,
                "--truth=" + self.truthtruth,
                self.sample_file_namesample_file_name,
            ]
            print("Running", cmd)
            subprocess.call(cmd)
 
            # Move training file to the right location
            if self.identifieridentifier.endswith(".xml"):
                tracking_data_dir_path = os.path.join(os.environ["BELLE2_LOCAL_DIR"], "tracking", "data")
                shutil.copy(self.identifieridentifier, tracking_data_dir_path)
 
        else:
            cmd = [
                "trackfindingcdc-classification-overview",
            ]
 
            if self.variablesvariables:
                cmd += ["-v"]
                cmd += self.variablesvariables
 
            if self.groupbygroupby is not None:
                cmd += ["-g"]
                if isinstance(self.groupbygroupby, str):
                    cmd += [self.groupbygroupby]
                else:
                    cmd += self.groupbygroupby
 
            if self.auxiliariesauxiliaries is not None:
                cmd += ["-a"]
                if isinstance(self.auxiliariesauxiliaries, str):
                    cmd += [self.auxiliariesauxiliaries]
                else:
                    cmd += self.auxiliariesauxiliaries
 
            cmd += [
                "--truth=" + self.truthtruth,
                self.sample_file_namesample_file_name,
            ]
            print("Running", cmd)
            subprocess.call(cmd)
            
            self.output_file_nameoutput_file_nameoutput_file_name = self.sample_file_namesample_file_name[:-len(".root")] + ".overview.root"
 
        super().postprocess()