Hype is a proof-of-concept deep learning library, where you can perform optimization on compositional machine learning systems of many components, even when such components themselves internally perform optimization.
It is developed by Atılım Güneş Baydin and Barak A. Pearlmutter, at the Brain and Computation Lab, National University of Ireland Maynooth.
This work is supported by Science Foundation Ireland grant 09/IN.1/I2637.
Please visit the project website for documentation and tutorials.
You can come and join the Gitter chat room, if you want to chat with us:
Hype is released under the MIT license.