Share Your Work - Assignment 2

Please share your work from Assignment 2 on this thread.

Share your Jupyter notebooks, blog posts, demo videos, etc. to get feedback on your work from the entire community, and you will also get to learn from all the other participants.

Reply to this thread to share what you’re working on. You can share the following:

  • Jupyter notebooks hosted on Jovian. Be sure to add a nice title and a helpful description of your work.
  • Blog posts or tutorials you have written as part of the assignments and course project (yes, you’ll be writing blog posts!)
  • Video demos or animations of your how your model is performing
  • Anything else you have created as part of this course, or otherwise.

Note: While commenting to others’ work, please be courteous, supportive and give constructive feedback to make this a positive learning environment for everyone.

This is my first notebook on “Linear Regression”, this is for learning purpose only and not for Assignment 2, please let me know, if anything is wrong or any advises to get better!
It will be very helpful to me! :star_struck:


This is simple and great, it would have been better if the accuracy or loss changes would have been printed say after every 1000 epochs and also a visual graph showing the improvement of the accuracy was shown.
The graph visualization might be helpful for you too as you will be able to see the trend in the model training and easily understand when the curve is flattening, this will help you to understand when to stop the training process. (over training on the same data may lead to overfitting).

1 Like

Thank you for checking my work out @birajde, it was simple as I tried to do it based on the things that I learned from 1st day lecture!:)) I have some doubts, so please help me out with the followings -

  • If I print after every 1000 epochs, won’t my project become length?

  • Even I thought of plotting it visually, but was clueless on how to do it. Any suggestion on that side?

  • If you think that your project will become lengthy if you print the accuracy or loss after every 1000 epochs do it after 2000 or 5000 epochs, it may seem lengthy initially but you will be well informed about the way your model is trained.(Selecting the correct choice of epochs is upto you and the model , if your model had only 100 200 epochs you could have printed the data after every 10 epochs but as there are a lot of epcohs for this model of yours choose some numbers suitable to you).
  • While you are printing the accuracy or loss after every 1000/2000/5000 epochs save the value of the accuracy/loss in a seperate list before printing. Thus you will get a list of accuracies/losses, you can easily make a graph out of it(Have you not watched the 2nd Week’s lecture yet? I think a similar graphs was covered in the 2nd Week’s Lecture).

No, I could not watch it till yet, will be completing by tonight and thank you for your help, will surely try working on those section which needs improvement! :smiley:

1 Like

Hi readers, hope you and your loved ones are all keeping well in terms of health. This is my work on the assignment of creating your first model(linear regression). Hope you all like it and I’m open to suggestions anytime. All the best to each one of my fellow learners.


Here is my submission, thanks to [SHREYAS CHATTERJEE] for guided me

1 Like

Hey all, I have completed my assignment.
Check here:

Need help? Just ping!


You are welcome mate

Guys, check this out and suggest me how should I improve my model.

1 Like

Hi everyone!!
I needed a favour regarding assignment 2.Can anyone suggest me what may be the value to stop the iterations.Like good val_loss??

Depends on what loss function you’ll use.

I would consider below 4000 for L1 to be a good approximation.

1 Like

You don’t have to make your own implementation of L1 loss to be honest :stuck_out_tongue: I’ve looked through your notebook and in previous versions you have used something like F.mse_loss. Pytorch actually provides an implementation of L1 loss as well: F.l1_loss.

Also, your learning rates seem to be a bit weird. I would start with 1e-3 or 1e-4 and then decrease.
1e-3 means 1 times 10 to the power of -3, so 1/(10^3)
1e-4 means 1 times 10 to the power of -4, so 1/(10^4)

What you have now are learning rates that actually increase instead of decreasing. It’s good idea to decrease the learning rates further into the training process.

1 Like

i know but I wanted to try littlr=e bit different.I have applied like the learning rate u have said but i see no such major difference

1 Like

Hello friends,
I have completed my second assignment , please give it a look and if you need guidance you can msg me on instagram my id is : tanujkumar2019

1 Like

Why you give northeast and northwest same encoding value. Is it mistake.

1 Like