The project is focussed on training a small Convolutional model (8 layers) to learn from the MNIST dataset and predict on similar content.
For training the model, you just need to run the S5.ipynb
file. The model is defined in the model.py
file, while the utils.py
file contains
the train and test functions for the model training and testing respectively along with the auxiliary function GetCorrectPredCount.
The model was trained for 5 epoch, with the metrics plotted below.