Getting started with Keras-team Keras

WML CE includes Conda packages for Keras-team Keras. Keras-team Keras has been configured to run with the Tensorflow back-end, and is also configured to operate with the Tensorflow Large Model Support (TFLMS).

The Keras-team Keras packages include the Keras-team Keras 2.2.5 version. See the Keras-team Keras Github repository for additional notes on Keras-team Keras.

Installing Keras

Install the keras-gpu Meta package to run with the Tensorflow GPU back-end:
```
conda install keras-gpu
```
Installing the keras-gpu package will also install the requisite tensorflow-gpu Conda packages and the keras-base requisite Conda package.
Install the keras Meta package to run with the Tensorflow Eigen back-end:
```
conda install keras
```
Installing the keras package also installs the requisite tensorflow-cpu Eigen Conda packages and the keras-base requisite Conda package.
To uninstall the keras packages, use the conda remove command to remove the individual packages:
```
conda remove keras-gpu
```
```
conda remove keras
```
```
conda remove keras-base
```

Keras examples

The Keras-team Keras package includes a set of examples. A script is provided to copy the sample content into a specified directory:

keras-install-samples <destination-directory>

Verifying the Keras installation

The Keras-team Keras package also includes a set of tests that can be used to verify the Keras-team Keras and Tensorflow GPU installation:

keras-test