Dropout not implemented in FCs

ArkinDharawat / DeepImageRanking

Implementation of `Learning Fine-grained Image Similarity with Deep Ranking` using ResNet-101 and implemented in Pytorch

23 stars 10 forks source link

Dropout not implemented in FCs #3

Open hchintada opened 4 years ago

hchintada commented 4 years ago

The paper says that a dropout layer with prob rate of 0.6 is implemented for all the fully connected layers in the network. Curious to know the reason why it is not added in the architecture given in the repo!

Do you suggest adding a dropout layer?

ArkinDharawat commented 4 years ago

Yes, feel free to add a dropout layer and create a pull request.