Releases: titu1994/Snapshot-Ensembles
Weights for CIFAR 100 (Wide ResNet - 16 - 4)
Contains the 6 weight files for the WRN-16-4 model, trained on the CIFAR 100 Dataset. Please download and extract weight files and place them in the weights folder.
Note : There are 4 versions. Choose the correct version for your backend [theano, tensorflow] and correct dim ordering [th, tf]
Optimized Ensemble Weight file:
Ensemble-weights-WRN-CIFAR100-16-4.json
Weight File Names :
WRN-CIFAR100-16-4-Best.h5
WRN-CIFAR100-16-4-1.h5
WRN-CIFAR100-16-4-2.h5
WRN-CIFAR100-16-4-3.h5
WRN-CIFAR100-16-4-4.h5
WRN-CIFAR100-16-4-5.h5
Weights for CIFAR 10 (Wide ResNet - 16 - 4)
Contains the 6 weight files for the WRN-16-4 model, trained on the CIFAR 10 Dataset. Please download and extract the weight files and place them in the weights
folder.
Note : There are 4 versions. Choose the correct version for your backend [theano, tensorflow] and correct dim ordering [th, tf]
Optimized Ensemble Weight file:
Ensemble-weights-WRN-CIFAR10-16-4.json
Weight File Names :
- WRN-CIFAR10-16-4-Best.h5
- WRN-CIFAR10-16-4-1.h5
- WRN-CIFAR10-16-4-2.h5
- WRN-CIFAR10-16-4-3.h5
- WRN-CIFAR10-16-4-4.h5
- WRN-CIFAR10-16-4-5.h5