Makes activity prediction of a guide-target pair on the basis of the outputs of two models: classification to determine whether a pair is active, and regression to determine if a pair is highly active (exceeds a threshold) [a44fe3d].
Includes example predictive models directly in the repository, serialized with TensorFlow's SavedModel format [e18a8a2].
Adds arguments to design.py that accept path to serialized classification and regression models, which can be the models included here or any other custom model [8ae1eee].
This PR makes several important changes:
design.pythat accept path to serialized classification and regression models, which can be the models included here or any other custom model [8ae1eee].