Phase I Deep CSV Training

cms-sw · Apr 11, 2017 · 2218370 · 2218370
1 parent 3611c52
commit 2218370
Show file tree

Hide file tree

Showing 6 changed files with 45 additions and 37 deletions.
diff --git a/PhysicsTools/PatAlgos/python/producersLayer1/jetProducer_cfi.py b/PhysicsTools/PatAlgos/python/producersLayer1/jetProducer_cfi.py
@@ -49,20 +49,18 @@
         # CTagging
         cms.InputTag('pfCombinedCvsLJetTags'),
         cms.InputTag('pfCombinedCvsBJetTags'),
-        # The following code is commented-out to avoid breaking any unit test
-        # waiting for a set of AOD RelVals which have the jet tags in the event content 
         # DeepFlavour
-        # cms.InputTag('pfDeepCSVJetTags:probb'),
-        # cms.InputTag('pfDeepCSVJetTags:probc'),
-        # cms.InputTag('pfDeepCSVJetTags:probudsg'),
-        # cms.InputTag('pfDeepCSVJetTags:probbb'),
-        # cms.InputTag('pfDeepCSVJetTags:probcc'),
+        cms.InputTag('pfDeepCSVJetTags:probb'),
+        cms.InputTag('pfDeepCSVJetTags:probc'),
+        cms.InputTag('pfDeepCSVJetTags:probudsg'),
+        cms.InputTag('pfDeepCSVJetTags:probbb'),
+        cms.InputTag('pfDeepCSVJetTags:probcc'),
         # DeepCMVA
-        # cms.InputTag('pfDeepCMVAJetTags:probb'),
-        # cms.InputTag('pfDeepCMVAJetTags:probc'),
-        # cms.InputTag('pfDeepCMVAJetTags:probudsg'),
-        # cms.InputTag('pfDeepCMVAJetTags:probbb'),
-        # cms.InputTag('pfDeepCMVAJetTags:probcc'),
+        cms.InputTag('pfDeepCMVAJetTags:probb'),
+        cms.InputTag('pfDeepCMVAJetTags:probc'),
+        cms.InputTag('pfDeepCMVAJetTags:probudsg'),
+        cms.InputTag('pfDeepCMVAJetTags:probbb'),
+        cms.InputTag('pfDeepCMVAJetTags:probcc'),
     ),
     # clone tag infos ATTENTION: these take lots of space!
     # usually the discriminators from the default algos
@@ -101,4 +99,14 @@
     resolutions     = cms.PSet()
 )
 
+from Configuration.Eras.Modifier_phase1Pixel_cff import phase1Pixel
+_phaseI_taggers = cms.VInputTag(
+   *[i for i in _patJets.discriminatorSources if i.value() != 'pfDeepCSVJetTags:probcc']
+)
+phase1Pixel.toModify(
+   _patJets, 
+   discriminatorSources = _phaseI_taggers
+)
+
+
 patJets = _patJets.clone()
diff --git a/PhysicsTools/PatAlgos/python/slimming/applySubstructure_cff.py b/PhysicsTools/PatAlgos/python/slimming/applySubstructure_cff.py
@@ -62,7 +62,7 @@ def applySubstructure( process ) :
                      jetSource = cms.InputTag('ak8PFJetsPuppi'),
                      algo= 'AK', rParam = 0.8,
                      jetCorrections = ('AK8PFPuppi', cms.vstring(['L2Relative', 'L3Absolute']), 'None'),
-                     btagDiscriminators = ([x.getModuleLabel() for x in patJetsDefault.discriminatorSources] + ['pfBoostedDoubleSecondaryVertexAK8BJetTags']),
+                     btagDiscriminators = ([x.value() for x in patJetsDefault.discriminatorSources] + ['pfBoostedDoubleSecondaryVertexAK8BJetTags']),
                      genJetCollection = cms.InputTag('slimmedGenJetsAK8')
                      )
     process.patJetsAK8Puppi.userData.userFloats.src = [] # start with empty list of user floats

diff --git a/RecoBTag/Combined/plugins/DeepFlavourJetTagsProducer.cc b/RecoBTag/Combined/plugins/DeepFlavourJetTagsProducer.cc
@@ -76,6 +76,8 @@ class DeepFlavourJetTagsProducer : public edm::stream::EDProducer<> {
 	// ----------member data ---------------------------
 	const edm::EDGetTokenT< INFOS > src_;
 	edm::FileInPath nnconfig_;
+	bool check_sv_for_defaults_;
+	bool mean_padding_;
 	lwt::LightweightNeuralNetwork *neural_network_;
 	lwt::ValueMap inputs_; //typedef of unordered_map<string, float>
 	vector<string> outputs_;
@@ -97,6 +99,8 @@ class DeepFlavourJetTagsProducer : public edm::stream::EDProducer<> {
 DeepFlavourJetTagsProducer::DeepFlavourJetTagsProducer(const edm::ParameterSet& iConfig) :
   src_( consumes< INFOS >(iConfig.getParameter<edm::InputTag>("src")) ),
 	nnconfig_(iConfig.getParameter<edm::FileInPath>("NNConfig")),
+	check_sv_for_defaults_(iConfig.getParameter<bool>("checkSVForDefaults")),
+	mean_padding_(iConfig.getParameter<bool>("meanPadding")),
 	neural_network_(NULL),
 	inputs_(),
 	outputs_(),
@@ -134,7 +138,8 @@ DeepFlavourJetTagsProducer::DeepFlavourJetTagsProducer(const edm::ParameterSet&
 																				 << ". Please check the spelling" <<  std::endl;
 		}
 		var.index = (tokens.size() == 2) ? stoi(tokens.at(1)) : -1;
-		var.default_value = -1*input.offset; //set default to -offset so that when scaling (val+offset)*scale the outcome is 0
+		var.default_value = (mean_padding_) ? 0. : -1*input.offset; //set default to -offset so that when scaling (val+offset)*scale the outcome is 0
+		//for mean padding it is set to zero so that undefined values are assigned -mean/scale
 
 		variables_.push_back(var);
 	}
@@ -184,9 +189,11 @@ DeepFlavourJetTagsProducer::produce(edm::Event& iEvent, const edm::EventSetup& i
 		TaggingVariableList vars = info.taggingVariables();
     //if there are no tracks there's no point in doing it
 		bool notracks = (vars.get(reco::btau::jetNSelectedTracks) == 0); 
+		bool novtx = (vars.get(reco::btau::jetNSecondaryVertices) == 0); 
+		bool defaulted = (check_sv_for_defaults_) ? (notracks && novtx) : notracks;
 		lwt::ValueMap nnout; //returned value
 
-		if(!notracks) {
+		if(!defaulted) {
 			for(auto& var : variables_) {
 				if(var.index >= 0){
 					std::vector<float> vals = vars.getList(var.id, false);
@@ -207,7 +214,7 @@ DeepFlavourJetTagsProducer::produce(edm::Event& iEvent, const edm::EventSetup& i
 
 		//dump the NN output(s)
 		for(size_t i=0; i<outputs_.size(); ++i) {
-			(*output_tags[i])[key] = (notracks) ? -1 : nnout[outputs_[i]];
+			(*output_tags[i])[key] = (defaulted) ? -1 : nnout[outputs_[i]];
 			//std::cout << i << ": " << nnout[outputs_[i]] << std::endl;
 		}
 	}

diff --git a/RecoBTag/Combined/python/deepFlavour_cff.py b/RecoBTag/Combined/python/deepFlavour_cff.py
@@ -52,8 +52,8 @@
 ##
 pfDeepFlavourTask = cms.Task(
     pfDeepCSVTagInfos,
-    ## pfDeepCMVATagInfos, #SKIP for the moment
+    pfDeepCMVATagInfos, #SKIP for the moment
     pfDeepCSVJetTags
-    ## , pfDeepCMVAJetTags
+    , pfDeepCMVAJetTags
 )
 pfDeepFlavour = cms.Sequence(pfDeepFlavourTask)
diff --git a/RecoBTag/Combined/python/pfDeepCSVJetTags_cfi.py b/RecoBTag/Combined/python/pfDeepCSVJetTags_cfi.py
@@ -3,5 +3,12 @@
 pfDeepCSVJetTags = cms.EDProducer(
 	'DeepFlavourJetTagsProducer',
 	src = cms.InputTag('pfDeepCSVTagInfos'),
+  checkSVForDefaults = cms.bool(False),
+  meanPadding = cms.bool(False),
 	NNConfig = cms.FileInPath('RecoBTag/Combined/data/DeepFlavourNoSL.json')
 	)
+
+from Configuration.Eras.Modifier_phase1Pixel_cff import phase1Pixel
+phase1Pixel.toModify(pfDeepCSVJetTags, NNConfig = cms.FileInPath('RecoBTag/Combined/data/DeepCSV_PhaseI.json'))
+phase1Pixel.toModify(pfDeepCSVJetTags, checkSVForDefaults = cms.bool(True))
+phase1Pixel.toModify(pfDeepCSVJetTags, meanPadding = cms.bool(True))
diff --git a/RecoBTag/Combined/scripts/format_deepflavour_json.sh b/RecoBTag/Combined/scripts/format_deepflavour_json.sh
@@ -1,23 +1,9 @@
 #! /bin/env bash
 
-sed -i 's|Jet_eta|jetEta|g' $1
-sed -i 's|Jet_pt|jetPt|g' $1
+#sed -i 's|jet_eta|jetEta|g' $1
+sed -i 's|jet_eta|jetAbsEta|g' $1
+sed -i 's|jet_pt|jetPt|g' $1
 sed -i 's|TagVarCSV_||g' $1
+sed -i 's|TagVarCSVTrk_||g' $1
 sed -i 's|prob_|prob|g' $1
-
-#bugfixes
-sed -i 's|jetNTracks|jetNSelectedTracks|g' $1
-sed -i 's|jetNSelectedTracksEtaRel|jetNTracksEtaRel|g' $1
-
-python <<EOF
-import json
-with open('$1') as infile:
-  jmap = json.loads(infile.read())
-
-for var in jmap['inputs']:
-  var['offset'] *= -1
-  var['scale'] = 1./var['scale']
-
-with open('$1', 'w') as out:
-  out.write(json.dumps(jmap, indent=2, separators = (',', ': ')))
-EOF
+sed -i 's|trackJetDistVal|trackJetDist|g' $1