Hi
I’m looking for pyTorch implementation of the paper:
From Pixels to Sentiment: Fine-tuning CNNs for Visual Sentiment Prediction (https://arxiv.org/pdf/1604.03489.pdf).
I am new in Deep Learning and it looks pretty complicated. However I need this network in Python. Anyone know if there is an implementation? I looked for but couldn’t find any
Apparently they use a network looking like Alexnet. Not very complicated and already implemented and trained in pytorch. Then finetuning and their classifier. Didn’t really look into the details 
@lelouedec You have a recommended implementation for AlexNet? Find several… but actually I’m interesting in the implementation of the finetuning… because AlexNet already exits
I think you can see the torch vision package for implementation detail
or this github on all kind of CNN model
Alexnet is implemented and available pretrained in pytorch. I let you look at tutorials explaining how to use it.
http://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html#sphx-glr-beginner-transfer-learning-tutorial-py
this tutorial covers everything you need to know with resnet, but it works the same way with pytorch
But they applied additional fune-tuning - how can I implement it?
firstable do you understand what finetuning is ?
Secondable if you do, the link I gave you introduce you to finetuning in pytorch, you will just need to adapte it to how the paper is doing it.
@lelouedec I thought finetuning is some additional added layers to find specific values of parameters? if this is not the case, please explain? Thanks
https://flyyufelix.github.io/2016/10/03/fine-tuning-in-keras-part1.html here a bit of reading for ya 
this plus the previous link should get you started on finetuning.