MNIST classification using pytorch

This repo provides an example of how to use Pytorch to train a simple CNN model to classify handwritten digits (MNIST) as is provided on the kaggle website https://www.kaggle.com/c/digit-recognizer.

It will also evaluate the kaggle MNIST test set and prepare a csv file for submission to this practice competition.

• Run the train.py file to train the model - a .pt model file will be saved after training.

• Run the evaluate.py file to load the saved .pt model file and evaluate on the kaggle test set - a csv file will be created which can be uploaded to the kaggle website.

Updated for pytorch 0.4.1