Final model for coding challenge for Fatima Fellowship

Trained the 'cifar100' dataset using the 'resnet18' pretrained model with on the fly random labelling images. In epoch-24 the train and validation accuracy is consecutively 0.991 and 0.906 and inference accuracy is 99.4% using 'cifar10' which is till now best accuracy have gotten for upside_down and normal image classification.

my_model_epoch_24.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0adc5e4eefc88fb91f3a43e5cae28086d08e772397d06dbf1c5ef8e250997d5
+size 134241877