learn2learn is a meta-learning library providing three levels of functionality for users. At a high level, there are many examples using meta-learning algorithms to train on a myriad of datasets/environments. At a mid level, it provides a functional interface for several popular meta-learning algorithms as well as a data loader to make it easier to import other data sets. At a low level, it provides extended functionality for modules.
A pip package is available, updated periodically. Use the command:
pip install -U learn2learn
For the most update-to-date version clone the repository and use:
pip install -e .
When installing from sources, make sure that Cython is installed:
pip install cython.
While learn2learn is actively used in current research projects, it is still in development. Breaking changes might occur.
To simplify the development process, the following commands can be executed from the cloned sources:
make build- Builds learn2learn in place.
make clean- Cleans previous installation.
make lint- Runs linting on the codebase.
make lint-examples- Runs linting on the examples.
make tests- Runs a light testing suite. (i.e. the Travis one)
make alltests- Runs an extensive testing suite. (much longer)
make docs- Builds the documentation and serves the website locally.
If you encounter a problem, feel free to an open an issue and we'll look into it.