Implementation of a paper

I’m looking for pyTorch implementation of the paper:
From Pixels to Sentiment: Fine-tuning CNNs for Visual Sentiment Prediction (
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 :confused:

@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.
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 here a bit of reading for ya :wink: this plus the previous link should get you started on finetuning.