A small helper library for building directed acyclic graphs in Theano

I got tired of writing shape tuples all over the place in my experiment code. Realizing (partly from dask - a wonderful library) that with the feature dimensionality of my data set and stored, all other shapes can be calculated on the fly, I set about trying to do this. With the help of the Theano core developers, I managed to cobble together a shape inference function and wrapped this library around it.

Easy prototyping of out-of-core recurrent models with conditional structures, simple monitoring via html plots, many wrapped examples and tests, and a focus on node/function based code instead of an object-oriented approach.

One other goal is to make sharing experiment code easy - the entire library is serialized during training, and tries to have absolute minimal dependencies besides numpy, scipy, and Theano. Eventually I would like to save a single file which has all the codepaths used in an experiment.

Philosophically, there is similarity to Lasagne (another great neural network library) but written with my own research goals in mind.

I (likely) won't be taking pull requests beyond bug fixes, though this may change. Feel free to fork and hack, and if something cool emerges we can bring it upstream.

If you use this library, I will likely break your code. Someday I hope to have enough examples, tests, and experiments to have the API solid, but for now consider this "bleeding edge" type development.

The typical install involves setting up a scientific Python environment using your preferred approach (I like Continuum Analytics Anaconda personally), plus the latest version of Theano.

Once this is done, clone this repo and run

python setup.py develop

Try running tests and examples to be sure install worked correctly.