Rudenyel initial commit

I am training a model for digit recognition in the Google Colab environment. https://colab.research.google.com/drive/1qnUQfxb-mOGOAW-y3xVVaCCcsiFDmNxB?usp=sharing The model is trained using MNIST test dataset - 60,000 images (28x28 pixels in size) for test data and 10,000 images for validation data. The model is trained using a simple Multilayer perceptron neural network - 28x28 input layers (based on the number of image pixels), 500 hidden layers, 10 output layers (based on the number of digits). The model is stored locally and used by a telegram bot to predict digits. The bot is based on the asynchronous aiogram library. Images are centered, cropped, made more contrasty, and brought to 28x28 size using the pillow library. The processed images are visualized using matplotlib library