Model save path: ./New_Models/bn_True_dataset_MNIST_epochs_200_lr_0.01_model_type_vgg19_rand_seed_265358_test_samples_None_train_samples_None_weight_decay_0.0005.pth.tar
Training Set:
Layer 0: Conv2d(1, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.09116753190755844
Inter Cos: 0.10967151820659637
Norm Quadratic Average: 23.567678451538086
Nearest Class Center Accuracy: 0.8079833333333334

Layer 1: Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.06391827762126923
Inter Cos: 0.07778043299913406
Norm Quadratic Average: 20.18315315246582
Nearest Class Center Accuracy: 0.8283

Layer 2: Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1040426567196846
Inter Cos: 0.09543313831090927
Norm Quadratic Average: 11.408917427062988
Nearest Class Center Accuracy: 0.8766166666666667

Layer 3: Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.09983926266431808
Inter Cos: 0.09113071858882904
Norm Quadratic Average: 12.09011173248291
Nearest Class Center Accuracy: 0.88625

Layer 4: Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.17198292911052704
Inter Cos: 0.1141885444521904
Norm Quadratic Average: 7.921565532684326
Nearest Class Center Accuracy: 0.9348833333333333

Layer 5: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.20425640046596527
Inter Cos: 0.12576907873153687
Norm Quadratic Average: 8.60227108001709
Nearest Class Center Accuracy: 0.9565833333333333

Layer 6: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.23160073161125183
Inter Cos: 0.12539105117321014
Norm Quadratic Average: 8.846695899963379
Nearest Class Center Accuracy: 0.9700833333333333

Layer 7: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.2593460977077484
Inter Cos: 0.12033002823591232
Norm Quadratic Average: 9.16576099395752
Nearest Class Center Accuracy: 0.9767833333333333

Layer 8: Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.3400363028049469
Inter Cos: 0.15207242965698242
Norm Quadratic Average: 6.637900352478027
Nearest Class Center Accuracy: 0.9933833333333333

Layer 9: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.5224810838699341
Inter Cos: 0.16907328367233276
Norm Quadratic Average: 7.035637855529785
Nearest Class Center Accuracy: 0.9980833333333333

Layer 10: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.6795654296875
Inter Cos: 0.12249115109443665
Norm Quadratic Average: 7.5821027755737305
Nearest Class Center Accuracy: 0.9994333333333333

Layer 11: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.7950581908226013
Inter Cos: 0.07937639206647873
Norm Quadratic Average: 7.744595050811768
Nearest Class Center Accuracy: 0.9999166666666667

Layer 12: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.8907110691070557
Inter Cos: 0.10978896170854568
Norm Quadratic Average: 6.7299346923828125
Nearest Class Center Accuracy: 0.9999833333333333

Layer 13: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9599587917327881
Inter Cos: 0.08680351078510284
Norm Quadratic Average: 4.204160213470459
Nearest Class Center Accuracy: 1.0

Layer 14: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9747746586799622
Inter Cos: 0.03477106988430023
Norm Quadratic Average: 4.223351955413818
Nearest Class Center Accuracy: 1.0

Layer 15: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9886077642440796
Inter Cos: -0.025213036686182022
Norm Quadratic Average: 4.267520904541016
Nearest Class Center Accuracy: 1.0

Layer 16: Linear(in_features=25088, out_features=4096, bias=True)
Linear Weight Norm: 14.677748680114746
Linear Weight Rank: 4031
Intra Cos: 0.9974503517150879
Inter Cos: -0.032965414226055145
Norm Quadratic Average: 43.27542495727539
Nearest Class Center Accuracy: 1.0

Layer 17: Linear(in_features=4096, out_features=4096, bias=True)
Linear Weight Norm: 6.268930912017822
Linear Weight Rank: 3667
Intra Cos: 0.9975867867469788
Inter Cos: 0.005634969100356102
Norm Quadratic Average: 27.009260177612305
Nearest Class Center Accuracy: 1.0

Layer 18: Linear(in_features=4096, out_features=10, bias=True)
Linear Weight Norm: 2.0679872035980225
Linear Weight Rank: 10
Intra Cos: 0.9970787167549133
Inter Cos: 0.030950110405683517
Norm Quadratic Average: 17.615676879882812
Nearest Class Center Accuracy: 1.0

Output Layer:
Intra Cos: 0.9996569156646729
Inter Cos: 0.11436617374420166
Norm Quadratic Average: 12.384773254394531
Nearest Class Center Accuracy: 1.0

Test Set:
Average Loss: 0.020537583645177073
Accuracy: 0.9959
NC1 Within Class Collapse: 0.08704261481761932
NC2 Equinorm: Features: 0.019982021301984787, Weights: 0.01121519599109888
NC2 Equiangle: Features: 0.08933053546481662, Weights: 0.06627458996242946
NC3 Self-Duality: 0.04755444824695587
NC4 NCC Mismatch: 0.00019999999999997797

Layer 0: Conv2d(1, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.10219582915306091
Inter Cos: 0.12048853933811188
Norm Quadratic Average: 23.595195770263672
Nearest Class Center Accuracy: 0.8229

Layer 1: Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.07160235196352005
Inter Cos: 0.07932940870523453
Norm Quadratic Average: 20.105989456176758
Nearest Class Center Accuracy: 0.8394

Layer 2: Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1139598861336708
Inter Cos: 0.09594735503196716
Norm Quadratic Average: 11.317767143249512
Nearest Class Center Accuracy: 0.8882

Layer 3: Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1091349869966507
Inter Cos: 0.09213259816169739
Norm Quadratic Average: 12.014081954956055
Nearest Class Center Accuracy: 0.8945

Layer 4: Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.18395541608333588
Inter Cos: 0.1172429621219635
Norm Quadratic Average: 7.866837024688721
Nearest Class Center Accuracy: 0.9418

Layer 5: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.21767497062683105
Inter Cos: 0.12371724843978882
Norm Quadratic Average: 8.541740417480469
Nearest Class Center Accuracy: 0.9624

Layer 6: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.24475634098052979
Inter Cos: 0.12320178747177124
Norm Quadratic Average: 8.792709350585938
Nearest Class Center Accuracy: 0.9716

Layer 7: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.27180126309394836
Inter Cos: 0.11800041049718857
Norm Quadratic Average: 9.114669799804688
Nearest Class Center Accuracy: 0.9764

Layer 8: Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.3504296839237213
Inter Cos: 0.1536225825548172
Norm Quadratic Average: 6.605232238769531
Nearest Class Center Accuracy: 0.9899

Layer 9: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.5294643640518188
Inter Cos: 0.16830672323703766
Norm Quadratic Average: 7.007223606109619
Nearest Class Center Accuracy: 0.9924

Layer 10: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.6827856302261353
Inter Cos: 0.11958521604537964
Norm Quadratic Average: 7.561549663543701
Nearest Class Center Accuracy: 0.9936

Layer 11: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.7945162057876587
Inter Cos: 0.07493969798088074
Norm Quadratic Average: 7.732698440551758
Nearest Class Center Accuracy: 0.9946

Layer 12: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.8832591772079468
Inter Cos: 0.10424290597438812
Norm Quadratic Average: 6.722103118896484
Nearest Class Center Accuracy: 0.9945

Layer 13: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9481420516967773
Inter Cos: 0.0858411118388176
Norm Quadratic Average: 4.19923734664917
Nearest Class Center Accuracy: 0.9953

Layer 14: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9597150683403015
Inter Cos: 0.03400959074497223
Norm Quadratic Average: 4.217881202697754
Nearest Class Center Accuracy: 0.9953

Layer 15: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9716895222663879
Inter Cos: -0.025898730382323265
Norm Quadratic Average: 4.260703086853027
Nearest Class Center Accuracy: 0.9958

Layer 16: Linear(in_features=25088, out_features=4096, bias=True)
Linear Weight Norm: 14.677748680114746
Linear Weight Rank: 4031
Intra Cos: 0.9799802899360657
Inter Cos: -0.03399655222892761
Norm Quadratic Average: 43.178062438964844
Nearest Class Center Accuracy: 0.9959

Layer 17: Linear(in_features=4096, out_features=4096, bias=True)
Linear Weight Norm: 6.268930912017822
Linear Weight Rank: 3667
Intra Cos: 0.9806986451148987
Inter Cos: 0.003844053717330098
Norm Quadratic Average: 26.948883056640625
Nearest Class Center Accuracy: 0.9959

Layer 18: Linear(in_features=4096, out_features=10, bias=True)
Linear Weight Norm: 2.0679872035980225
Linear Weight Rank: 10
Intra Cos: 0.9802919626235962
Inter Cos: 0.03245209902524948
Norm Quadratic Average: 17.57838249206543
Nearest Class Center Accuracy: 0.996

Output Layer:
Intra Cos: 0.985925018787384
Inter Cos: 0.12320785969495773
Norm Quadratic Average: 12.354808807373047
Nearest Class Center Accuracy: 0.996

