Logistic regression (Jupiter, pytorch)

Start with this code ; https://drive.google.com/file/d/1I05nfzpRfdZPhcBGNUP5q2Tf5a9u2x6G/view?usp=sharing
dataset: https://drive.google.com/file/d/1AERCbcGrbp66NUrQLW-HboCAyXdHvo_K/view?usp=sharing
Extend/improve the code to do the following:

Graph train/val loss and accuracy per epoch (two separate graphs). Use a legend to indicate which line is train and which is val.
Make sure you do not cause overfitting (as indicated by your plots).
Improve the accuracy so the error on the test dataset (as computed by sklearn, we’re not talking about loss here) is <17%.
Show a confusion matrix of the prediction errors

Sample Solution