Model save path: /content/drive/MyDrive/Neural Collapse/New_Models/bn_True_dataset_MNIST_epochs_100_lr_0.001_model_type_vgg11_weight_decay_0.003.pth.tar
Training Set:
Layer 0: Conv2d(1, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.08946064114570618
Inter Cos: 0.11311884969472885
Norm Quadratic Average: 23.532936096191406
Nearest Class Center Accuracy: 0.81175

Layer 1: Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1052183136343956
Inter Cos: 0.12060689926147461
Norm Quadratic Average: 74.80557250976562
Nearest Class Center Accuracy: 0.835875

Layer 2: Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.14624656736850739
Inter Cos: 0.13679851591587067
Norm Quadratic Average: 52.684783935546875
Nearest Class Center Accuracy: 0.8485

Layer 3: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.14562858641147614
Inter Cos: 0.12147609889507294
Norm Quadratic Average: 49.22980499267578
Nearest Class Center Accuracy: 0.870375

Layer 4: Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.16358745098114014
Inter Cos: 0.09617889672517776
Norm Quadratic Average: 29.949058532714844
Nearest Class Center Accuracy: 0.910125

Layer 5: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1696680784225464
Inter Cos: 0.08854714035987854
Norm Quadratic Average: 31.09562110900879
Nearest Class Center Accuracy: 0.935625

Layer 6: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1897997409105301
Inter Cos: 0.07305984944105148
Norm Quadratic Average: 21.09568214416504
Nearest Class Center Accuracy: 0.97575

Layer 7: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.28334105014801025
Inter Cos: 0.08677425235509872
Norm Quadratic Average: 16.51606559753418
Nearest Class Center Accuracy: 0.997625

Layer 8: Linear(in_features=25088, out_features=4096, bias=True)
Linear Weight Norm: 87.78446197509766
Linear Weight Rank: 4031
Intra Cos: 0.5155999660491943
Inter Cos: 0.1181417927145958
Norm Quadratic Average: 107.06385040283203
Nearest Class Center Accuracy: 1.0

Layer 9: Linear(in_features=4096, out_features=4096, bias=True)
Linear Weight Norm: 35.483428955078125
Linear Weight Rank: 3671
Intra Cos: 0.6692607402801514
Inter Cos: 0.14586052298545837
Norm Quadratic Average: 54.694095611572266
Nearest Class Center Accuracy: 1.0

Layer 10: Linear(in_features=4096, out_features=10, bias=True)
Linear Weight Norm: 2.055258274078369
Linear Weight Rank: 10
Intra Cos: 0.7915423512458801
Inter Cos: 0.17766593396663666
Norm Quadratic Average: 33.300559997558594
Nearest Class Center Accuracy: 1.0

Output Layer:
Intra Cos: 0.9149855971336365
Inter Cos: 0.27136990427970886
Norm Quadratic Average: 17.460006713867188
Nearest Class Center Accuracy: 1.0

Test Set:
Average Loss: 0.08259337276220322
Accuracy: 0.975
NC1 Within Class Collapse: 1.5999176502227783
NC2 Equinorm: Features: 0.07687146961688995, Weights: 0.01680051162838936
NC2 Equiangle: Features: 0.20300354427761502, Weights: 0.08705779711405436
NC3 Self-Duality: 0.5821866393089294
NC4 NCC Mismatch: 0.007499999999999951

Layer 0: Conv2d(1, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.10133810341358185
Inter Cos: 0.11957792192697525
Norm Quadratic Average: 22.82430076599121
Nearest Class Center Accuracy: 0.809

Layer 1: Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.12080249935388565
Inter Cos: 0.13038599491119385
Norm Quadratic Average: 73.61656951904297
Nearest Class Center Accuracy: 0.826

Layer 2: Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.14318768680095673
Inter Cos: 0.15494389832019806
Norm Quadratic Average: 51.857418060302734
Nearest Class Center Accuracy: 0.845

Layer 3: Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.14010880887508392
Inter Cos: 0.14055341482162476
Norm Quadratic Average: 48.56572723388672
Nearest Class Center Accuracy: 0.8645

Layer 4: Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.15366919338703156
Inter Cos: 0.11625626683235168
Norm Quadratic Average: 29.731048583984375
Nearest Class Center Accuracy: 0.9005

Layer 5: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1567222774028778
Inter Cos: 0.11055798828601837
Norm Quadratic Average: 30.92388153076172
Nearest Class Center Accuracy: 0.923

Layer 6: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.1867426037788391
Inter Cos: 0.08844908326864243
Norm Quadratic Average: 20.933734893798828
Nearest Class Center Accuracy: 0.9535

Layer 7: Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
Intra Cos: 0.25909897685050964
Inter Cos: 0.09803890436887741
Norm Quadratic Average: 16.278554916381836
Nearest Class Center Accuracy: 0.9695

Layer 8: Linear(in_features=25088, out_features=4096, bias=True)
Linear Weight Norm: 87.78446197509766
Linear Weight Rank: 4031
Intra Cos: 0.42939063906669617
Inter Cos: 0.12130488455295563
Norm Quadratic Average: 104.47547912597656
Nearest Class Center Accuracy: 0.975

Layer 9: Linear(in_features=4096, out_features=4096, bias=True)
Linear Weight Norm: 35.483428955078125
Linear Weight Rank: 3671
Intra Cos: 0.5668962001800537
Inter Cos: 0.15311065316200256
Norm Quadratic Average: 53.06145477294922
Nearest Class Center Accuracy: 0.9765

Layer 10: Linear(in_features=4096, out_features=10, bias=True)
Linear Weight Norm: 2.055258274078369
Linear Weight Rank: 10
Intra Cos: 0.6717240214347839
Inter Cos: 0.18992359936237335
Norm Quadratic Average: 32.192161560058594
Nearest Class Center Accuracy: 0.9755

Output Layer:
Intra Cos: 0.79982590675354
Inter Cos: 0.27497032284736633
Norm Quadratic Average: 16.784502029418945
Nearest Class Center Accuracy: 0.9765

