Reorganize ParticleNet models and add README.

Author: hqucms

ParticleNet/V00/ParticleNet-0000.params …tAK8/General/V00/ParticleNet-0000.paramsParticleNet/V00/ParticleNet-0000.params renamed to ParticleNetAK8/General/V00/ParticleNet-0000.params ParticleNet/V00/ParticleNet-0000.params renamed to ParticleNetAK8/General/V00/ParticleNet-0000.params

@@ -0,0 +1,26 @@
+# ParticleNetAK8-General-V00
+
+This folder contains the MXNet models for the ParticleNet tagger designed for identifying Top/W/Z/H and their decay modes. No mass decorrelation is performed. The model is trained on AK8Puppi jets using `RunIIFall17MiniAODv2` samples. 
+
+The output classes of the DNN are:
+ - probTbcq
+ - probTbqq
+ - probTbc
+ - probTbq
+ - probTbel
+ - probTbmu
+ - probTbta
+ - probWcq
+ - probWqq
+ - probZbb
+ - probZcc
+ - probZqq
+ - probHbb
+ - probHcc
+ - probHqqqq
+ - probQCDbb
+ - probQCDcc
+ - probQCDb
+ - probQCDc
+ - probQCDothers
+

ParticleNet/V00/ParticleNet-symbol.json …tAK8/General/V00/ParticleNet-symbol.jsonParticleNet/V00/ParticleNet-symbol.json renamed to ParticleNetAK8/General/V00/ParticleNet-symbol.json ParticleNet/V00/ParticleNet-symbol.json renamed to ParticleNetAK8/General/V00/ParticleNet-symbol.json

@@ -0,0 +1,14 @@
+# ParticleNetAK8-MD-2prong-V00
+
+This folder contains the MXNet models for the ParticleNet tagger designed for mass-independent 2-prong jet tagging (i.e., X->bb, X->cc, X->qq). The mass decorrelation is achieved by using a dedicated signal sample w/ a flat mass distribution (`/BulkGravitonToHHTo4Q_MX-600to6000_MH-15to250_part*_TuneCP5_13TeV-madgraph_pythia8/RunIIFall17MiniAODv2-PU2017_12Apr2018_multigridpack_94X_mc2017_realistic_v14-v1/MINIAODSIM`). The model is trained on AK8Puppi jets using `RunIIFall17MiniAODv2` samples. 
+
+The output classes of the DNN are:
+ - probXbb
+ - probXcc
+ - probXqq
+ - probQCDbb
+ - probQCDcc
+ - probQCDb
+ - probQCDc
+ - probQCDothers
+