lccurious / machine-learning-template

A Machine Learning Template Project

This is a machine learning template project for generic training, validation and testing.

Project Structure

machine-learning-template
├─data
├─experiments
├─lib
├─notebooks
├─tests
└─tools

• Project is mainly considered to be structured, where the core functional about config, models, core, loss etc. are stored in example_lib directory.

• And the tests directory are mainly consists of some test case of functional components. Generate the random inputs and test if there is any bug.

• The notebooks directory consists of some visulization results and early developement of some functional modules.

• The experiments directory consists of custom defined .yaml file for ablation studies.

• The tools directory consists of custom training and testing scripts.

• The data directory consists of datasets for training, some subdirectory can be created by symbol link: ln -s /path/to/target/dataset dataset_name

IDE Tips

If use VS Code, put following snippets in your .vscode/setting.json:

{
    "python.autoComplete.extraPaths": [
        "${workspaceFolder}/*"
    ],
    "python.analysis.extraPaths": [
        "${workspaceFolder}/*"
    ]
}

If use Pycharm, just mark related directories as source folders.

Create Requirements

Install the pipreqs, and use this tools to scan your project, and generate the requirements.txt.