This project was rectuitment task, for 2020 Allegro summer internship in data science field. Despite not getting internship, I consider it to be very enriching experience, and I'm satisfied with my results.
The task was to write (without external ML libraries) and train on MNIST data set, a model that would predict if hand wrote digit is primary or complex. It had to implement Stochastic gradient descent with momentum on logistic regresion.
A complete overview (in Polish) of results is included in sprawozdanie.pdf.