Contents

App structure

When creating an app with the CLI, you will download the current GitHub folder with the following structure.

Important

The main class must be named App in app.py ins the src folder.

├── README.md # explains how to create an app
├── __init__.py
├── checkpoint
│   ├── __init__.py
│   └── model.pt: Optionnal # optional: model to be used in the app must be placed there
├── examples
│   ├── app_with_checkpoint.py # example: app example with checkpoint
│   └── torch_classifier.py # example: show how to train a neural network with pre-trained embeddings
└── src
    ├── DESC.md # Desciption file of the application, feel free to put a maximum of information.
    ├── __init__.py
    ├── app.py # main application script. The main class must be named App.
    └── model.py: Optional # file to register the models you use in app.py.
    └── tags.json # file to register the tags on the hub.

Tags

For your app to be visible and well documented, tags should be filled to precise at least the tasks section. It will be really useful to retrieve it from deepchain hub.

  • tasks

  • librairies

  • embeddings

  • datasets

  • device

Important

If you want your app to benefit from deepchain’ GPU, set device to “gpu” in tags. It will run on “cpu” by default.

Special method in apps

Every apps inherit from the DeepChainApp class that provides some special method. This method is useful for loading extra files in your app.

  • Use method get_checkpoint_path(__file__) to get the absolute path of the file in the checkpoint folder.

  • Use method get_filepath(__file__,file) to get the absolute path of the file in the src folder.

DeepchainHub

All the applications that have been released publicly can be found on the hub as described in this section

How it works CLI