Deep Glue Usage Guide
Getting started with Deep Glue.
Installation
In the virtual environment in which you have torch installed:
pip install deepglue
Or clone repo and do editable install with:
pip install -e .
Usage
Import and use:
import deepglue as dg
Then all functions will be available. For instance to train a model for one epoch:
dg.train_one_epoch(model, data_loader, loss_function, optimizer)
To inspect the different functions that are available, and their usage, see the API page.
Demo notebook
Our demo notebook walks you through deepglue's core features with a hands-on example. Highlights include:
- Setting up the project structure.
- Preparing datasets and data loaders.
- Defining and training a network.
- Visualizating feature clusters in the trained network.
More demos will be forthcoming as we continue adding features do deepglue (see the roadmap).