Share Your Work - Assignment 3

Hi everyone,

Please find my solution for assignment 3. The best test accuracy I achieved is 52.4% with four layer feedforward NN (2048, 1024, 512, 10).

Thank you.

Best.

Hi, Guys After Hard Efforts I am able to complete 3rd Assignment click_here

Hello Everybody!
Please find attached my 3rd assignment here: https://jovian.ai/nataliia/03-cifar10-feedforward/v/37
Trying many different options I finally got the test accuracy ~59% and test loss ~1.31 .
Thank you and best regards!

The highest i have got,

Hello Guys,

Here is my third assignment 03-cifar10-feedforward

Thanks,
Vishvaraj Dhanawade

Hello all,
Sorry,Am very late to the submission .hope i will be considered .Please check my notebook on ASSIGNMENT 3 HERE .
Provide suggestions as well mt assignments are considered are not?

I forgot to share my assignment

Here is my solution for assignment 3

Hi Friends,

I am sharing my work on Assignment 3 here.

Thanks,
Chaitanya

Hii all,
here is my assignmet 3` .

Hi Team,

Here is mu submission for Task3.

Thank you.

Regards,
Janardan Chavan

Hi all,

Here is my assignment https://jovian.ai/dave-raghav/03-cifar10-feedforward. I’ve achieved about 51% accuracy.

Hello Everyone,

My assignment is here - https://jovian.ai/rrayanna/03-cifar10-feedforward. Achieved about 48% accuracy.

Please share your comments.

Updated notebook link: https://jovian.ai/shnem/03-cifar10-feedforward

i keep loosing my work on jovian.ai for some reason.

I tried my best in this notebook and achieved accuracy of 52%.
Just take look :diamonds: https://jovian.ai/ap-101/03-cifar10-feedforward :slight_smile:

Hey guys, I have started this assignment a bit late but needed some help.
I have shifted the model and the and also the inputs to the GPU but still while training on Kaggle I only see my CPU being utilised and not the GPU.
here is the the link to my notebook.

Hi @sahaajchawla I went through your code and it is correct. My guess is that kaggle provide the GPU as support to the CPU, so as long as the CPU can handle it you won’t be using the GPU (this is my assumption). I had the same issue, but when I started training more complex models the GPU got activated. One other thing you could do is trying to play with hyper-parameters like the batch size, because they might affect whether the GPU is used or not. For instance, I tried to increase the batch size from 128 to 256 and I saw that the GPU started working a bit. Good luck, and if you have further doubts do not hesitate to ask for advice! (.

I would like to express gratitude to jaisejohnson06 for feedback and help during my completion of the course.

Thanks a lot @micheledifazio I also tried and figured out that i can increase the batch size till it does not affect my training and that will utilize the GPU as well.