Model save path: ./New_Models/bn_False_dataset_MNIST_epochs_200_lr_0.01_model_type_vgg19_rand_seed_338327_test_samples_None_train_samples_None_weight_decay_0.0003.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: 0.10967152565717697
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: 0.0896248146891594
Inter Cos: 0.10771230608224869
Norm Quadratic Average: 59.12957000732422
Nearest Class Center Accuracy: 0.8140833333333334

Layer 2: Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.12353434413671494
Inter Cos: 0.13506245613098145
Norm Quadratic Average: 64.32549285888672
Nearest Class Center Accuracy: 0.8397166666666667

Layer 3: Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.12881433963775635
Inter Cos: 0.13874536752700806
Norm Quadratic Average: 87.55333709716797
Nearest Class Center Accuracy: 0.84795

Layer 4: Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.21387864649295807
Inter Cos: 0.16922007501125336
Norm Quadratic Average: 59.572044372558594
Nearest Class Center Accuracy: 0.90705

Layer 5: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.24863937497138977
Inter Cos: 0.16925574839115143
Norm Quadratic Average: 50.48351287841797
Nearest Class Center Accuracy: 0.9358333333333333

Layer 6: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.28489887714385986
Inter Cos: 0.16318689286708832
Norm Quadratic Average: 38.07208251953125
Nearest Class Center Accuracy: 0.9548833333333333

Layer 7: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.3233965039253235
Inter Cos: 0.18418018519878387
Norm Quadratic Average: 28.361087799072266
Nearest Class Center Accuracy: 0.965

Layer 8: Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.3682086169719696
Inter Cos: 0.17247265577316284
Norm Quadratic Average: 12.772793769836426
Nearest Class Center Accuracy: 0.9819833333333333

Layer 9: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.521498441696167
Inter Cos: 0.2489425241947174
Norm Quadratic Average: 9.853677749633789
Nearest Class Center Accuracy: 0.99115

Layer 10: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.5981208086013794
Inter Cos: 0.30580753087997437
Norm Quadratic Average: 10.328861236572266
Nearest Class Center Accuracy: 0.9945666666666667

Layer 11: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.6287862062454224
Inter Cos: 0.3422253131866455
Norm Quadratic Average: 11.41208267211914
Nearest Class Center Accuracy: 0.99555

Layer 12: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.651206374168396
Inter Cos: 0.31696945428848267
Norm Quadratic Average: 7.719464302062988
Nearest Class Center Accuracy: 0.9955166666666667

Layer 13: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.8692183494567871
Inter Cos: 0.3460405170917511
Norm Quadratic Average: 6.702599048614502
Nearest Class Center Accuracy: 0.9978

Layer 14: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9066048264503479
Inter Cos: 0.32410484552383423
Norm Quadratic Average: 6.469186782836914
Nearest Class Center Accuracy: 0.9982666666666666

Layer 15: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9155214428901672
Inter Cos: 0.3380146920681
Norm Quadratic Average: 6.112040042877197
Nearest Class Center Accuracy: 0.9987333333333334

Layer 16: Linear(in_features=25088, out_features=4096, bias=True)
Linear Weight Norm: 31.699138641357422
Linear Weight Rank: 4031
Intra Cos: 0.9278252720832825
Inter Cos: 0.2807592749595642
Norm Quadratic Average: 34.667884826660156
Nearest Class Center Accuracy: 0.9992333333333333

Layer 17: Linear(in_features=4096, out_features=4096, bias=True)
Linear Weight Norm: 13.025879859924316
Linear Weight Rank: 3669
Intra Cos: 0.9367912411689758
Inter Cos: 0.26725420355796814
Norm Quadratic Average: 29.45650291442871
Nearest Class Center Accuracy: 0.9995

Layer 18: Linear(in_features=4096, out_features=10, bias=True)
Linear Weight Norm: 2.5532493591308594
Linear Weight Rank: 10
Intra Cos: 0.9403937458992004
Inter Cos: 0.25077933073043823
Norm Quadratic Average: 25.402883529663086
Nearest Class Center Accuracy: 0.99975

Output Layer:
Intra Cos: 0.9700514078140259
Inter Cos: 0.31893324851989746
Norm Quadratic Average: 23.992904663085938
Nearest Class Center Accuracy: 0.9999833333333333

Test Set:
Average Loss: 0.02141454922081157
Accuracy: 0.9952
NC1 Within Class Collapse: 0.5454623699188232
NC2 Equinorm: Features: 0.10100898891687393, Weights: 0.029036592692136765
NC2 Equiangle: Features: 0.25780826144748265, Weights: 0.12637644873725043
NC3 Self-Duality: 0.18243037164211273
NC4 NCC Mismatch: 0.0022999999999999687

Layer 0: Conv2d(1, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.10219582170248032
Inter Cos: 0.12048853188753128
Norm Quadratic Average: 23.595197677612305
Nearest Class Center Accuracy: 0.8229

Layer 1: Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.10045616328716278
Inter Cos: 0.11771233379840851
Norm Quadratic Average: 59.2067756652832
Nearest Class Center Accuracy: 0.8256

Layer 2: Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.13741819560527802
Inter Cos: 0.14447662234306335
Norm Quadratic Average: 64.15386199951172
Nearest Class Center Accuracy: 0.8557

Layer 3: Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.14231644570827484
Inter Cos: 0.15086953341960907
Norm Quadratic Average: 87.44480895996094
Nearest Class Center Accuracy: 0.8597

Layer 4: Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.23169174790382385
Inter Cos: 0.1836627870798111
Norm Quadratic Average: 59.483360290527344
Nearest Class Center Accuracy: 0.9177

Layer 5: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.26633256673812866
Inter Cos: 0.18330971896648407
Norm Quadratic Average: 50.44790267944336
Nearest Class Center Accuracy: 0.9451

Layer 6: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.30300065875053406
Inter Cos: 0.1766296774148941
Norm Quadratic Average: 38.06915283203125
Nearest Class Center Accuracy: 0.9612

Layer 7: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.34203171730041504
Inter Cos: 0.19919903576374054
Norm Quadratic Average: 28.387237548828125
Nearest Class Center Accuracy: 0.9682

Layer 8: Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.38488441705703735
Inter Cos: 0.1849411129951477
Norm Quadratic Average: 12.800424575805664
Nearest Class Center Accuracy: 0.982

Layer 9: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.533257246017456
Inter Cos: 0.26253125071525574
Norm Quadratic Average: 9.892762184143066
Nearest Class Center Accuracy: 0.9878

Layer 10: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.6068886518478394
Inter Cos: 0.3228101134300232
Norm Quadratic Average: 10.386011123657227
Nearest Class Center Accuracy: 0.9894

Layer 11: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.6349596977233887
Inter Cos: 0.36055245995521545
Norm Quadratic Average: 11.480772972106934
Nearest Class Center Accuracy: 0.9904

Layer 12: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.6509546637535095
Inter Cos: 0.3295416533946991
Norm Quadratic Average: 7.764121055603027
Nearest Class Center Accuracy: 0.9899

Layer 13: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.8679761290550232
Inter Cos: 0.35743579268455505
Norm Quadratic Average: 6.753726005554199
Nearest Class Center Accuracy: 0.9914

Layer 14: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9131948351860046
Inter Cos: 0.33559733629226685
Norm Quadratic Average: 6.522135257720947
Nearest Class Center Accuracy: 0.9919

Layer 15: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.9214889407157898
Inter Cos: 0.34179213643074036
Norm Quadratic Average: 6.16023588180542
Nearest Class Center Accuracy: 0.9927

Layer 16: Linear(in_features=25088, out_features=4096, bias=True)
Linear Weight Norm: 31.699138641357422
Linear Weight Rank: 4031
Intra Cos: 0.9329016208648682
Inter Cos: 0.29198434948921204
Norm Quadratic Average: 34.91617202758789
Nearest Class Center Accuracy: 0.9932

Layer 17: Linear(in_features=4096, out_features=4096, bias=True)
Linear Weight Norm: 13.025879859924316
Linear Weight Rank: 3669
Intra Cos: 0.9412002563476562
Inter Cos: 0.2787400484085083
Norm Quadratic Average: 29.66878890991211
Nearest Class Center Accuracy: 0.9936

Layer 18: Linear(in_features=4096, out_features=10, bias=True)
Linear Weight Norm: 2.5532493591308594
Linear Weight Rank: 10
Intra Cos: 0.9443149566650391
Inter Cos: 0.2623401880264282
Norm Quadratic Average: 25.588409423828125
Nearest Class Center Accuracy: 0.9942

Output Layer:
Intra Cos: 0.9670073390007019
Inter Cos: 0.32286950945854187
Norm Quadratic Average: 24.17144012451172
Nearest Class Center Accuracy: 0.9945

