PyTorch MNSIT Handwritten Digits Demo

This project showcases PyTorch concepts by building a model for image classification on the classic MNSIT Handwritten Digits data set.

A Dense Neural Network is chosen by performing Random Search on different model and training hyper parameters.

To run the code

• Install project requirements:

  pip install -r requirements.txt

• Type the following command to run the code: python -m src

(The code was tested with Python 3.5)

Package Structure

• data: Contains classes for loading data sets including Datasets, Transforms, DataSamplers and DataLoaders.

• models: Contains different models

• training: Contains different components for building and training models

  • scoring_funcs: Contains functions used to measure a model's performance
  • tuning: Contains components used for model tuning
    • parameters_domain: Represents the domain where the model parameters live.
    • scorers: Measures the model's performance on a given data set.
    • search: Represents a searching algorithm for tuning a model.