Assignment 5 - Course Project

The error is size mismatch or cuda out of memory? Could you share reference to your Notebook or image of the error?

Here it is: https://jovian.ml/aysemineevren/course-project-flower-classification

Thank you very much

@aysemineevren
This is a size mismatch error and not “Cuda not available available error” , in the classifier layer(last layer) the nn.linear function’s in size is given 512 while the in size expected to be 25088 so either change the nn.Linear function as nn.Linear(25088,num_classes) .

1 Like

A post was merged into an existing topic: Share your work : July 6 - July 12

Yes you are right, it was memory before but not now. Thank you, it worked!

2 posts were merged into an existing topic: Share your work : July 6 - July 12

I didn’t know where to ask this, so putting this up here. I was just curious about when we will be receiving our certificates?

6 Likes

A post was merged into an existing topic: Share your work : July 6 - July 12

A post was merged into an existing topic: Share your work : July 6 - July 12

Hey guys,
As we have completed the courese i would like to deep dive into this field and to try new deep learning libraries.
Fastai library is amazing and built upon pytorch. Hopefully, we can understand it and do better things.
If anyone interested we can make a study group, study together and help each other in advancement of skill…
We can discuss the possibilities.
mail me at :- kaushikabhishek441@gmail.com
Thanks and regards
I don’t know where to put this so i did it here.

A post was merged into an existing topic: Share your work : July 6 - July 12

Cool. I’m glad that it worked for you :slight_smile:

1 Like

I know this might not be the appropriate thread (we don’t have a thread for competitions yet), but can anyone tell how someone can approach the OSIC Pulmonary Fibrosis competition using PyTorch?

Here’s my work so far, I just want to integrate PyTorch with it:

https://www.kaggle.com/aadhavvignesh/visualize-scan-by-video-dicom-files-metadata-eda

It would be great if you can suggest how to build a model for this case, everyone can get a opportunity to extend their acquired knowledge on real-world situations like this competition. :slight_smile:

Finally finished my blog for the project on Zoo Classification using CNN. Do check it out, would like your views and suggestions.

Hi All,
Does any one of you received the course certificate?

Hey @ms-krajesh , Yes I have received my course certificate nearly 12 days ago and I think many of us have also received it during that time .

Hello. I was lack of time. And now finely Ive finished my work. It takes me about 60 more hours to finish.
Its Segmentation CNN based on U-net model and resnet 50.
https://www.kaggle.com/novikovmikhail/asigment5-zero-to-gans-novikov-ms
__results___42_1

My thanks to Aakash NS due to introduction in torch. It works!!!

p.s. I already get a certificate from you and here is a blog post :slight_smile:
https://medium.com/@Mikhail_SN/my-intro-in-neural-networks-286926cd191a?sk=6c2f99df595b3f9bd3948f184c1293a1

1 Like

how can i download the cifar100 dataset.

Hi @jgaurav2001 You can download it from torchvision.datasets. Just like this:
ds = torchvision.datasets.CIFAR100(root=“your_root_dir”, train=True, download=True)