NEWS.md
In kerasR: R Interface to the Keras Deep Learning Library

kerasR 0.7.0 (2017-06-19)

The test suite is built out to include coverage for all available user-facing functions. We also now explicitly check and warn users if they attempt to call a function without first installing keras. Explicit pip commands are given for accomplishing this.

We also now print the path to python that reticulate is currently linked to. The reticulate package cannot be restarted and because it will silently ignore incorrect paths, this has lead to a lot of bug reports that boil down to users not selecting the correct version of python. Hopefully this change will make it more obvious that this is the issue.

kerasR 0.6.1 (2017-06-01)

This version adding a testing suite for of the core functions in the library and conforming with the goodpractice::gp() recommendations. Also adds hooks to TravisCL and AppVeyor for integrated testing.

kerasR 0.5.0 (2017-04-26)

Adding the following functions to make it easier to initalize and check that keras is properly installed; these also make it possible to restart the python engine if needed.

keras_available()
keras_init()

kerasR 0.4.1 (2017-03-20)

This is the inital working version that was publicly pushed to GitHub. Older versions only ran locally on the development machine. Currently all of the layers in the keras library other than the functional layers are wrapped and exported.