Interface to 'Keras' < https://keras.io>, a high-level neural networks 'API'. 'Keras' was developed with a focus on enabling fast experimentation, supports both convolution based networks and recurrent networks (as well as combinations of the two), and runs seamlessly on both 'CPU' and 'GPU' devices.
Models saved via
export_savedmodel() that make use of learning phases can
now be exported without having to manually reload the original model.
Ensure that models saved via
export_savedmodel() can be served from CloudML
Run image data generators with R preprocessing functions on the main thread
Return R list from
Various fixes for
theme_bw option to plot method for training history
Support TF Dataset objects as generators for
use_backend() functions as alternative to
KERAS_BACKEND environment variables.
Added R wrappers for Keras backend functions (e.g.
Use 1-based axis for
Fix issue with printing training history after early stopping.
Experimental support for using the PlaidML backend.
Correct handling for R functions specified in
Automatically provide name to loss function during compile (enables save/load of models with custom loss function)
keras.fit_verbose option (defaults to 1)
keras_array() on the results of generator functions.
steps_per_epoch is passed as an integer
evaluate() generic from tensorflow package
NULL when converting R arrays to Keras friendly arrays
sample_weight is passed to
fit() as an array.
Accept single function as
metrics argument to
input_shape argument to applications to integer
Allow Keras models to be composable within model pipelines
export_savedmodel() generic from TensorFlow package
Convert R arrays to row-major before image preprocessing
tensorflow.keras for tensorflow implementation (TF v1.4)
application_inception_resnet_v2() pre-trained model
recurrent layers backed by CuDNN)
interpolation parameter to
Fix for progress bar output in Keras >= 2.0.9
implementation argument from recurrent layers
Support for passing generators for validation data in
Accept single integer arguments for kernel sizes
Add standard layer arguments to
image_array_save() for 3D image arrays.
Allow custom layers and lambda layers to accept list parameters.
add_loss() function for custom layers
use_session_with_seed() function that establishes a random seed for the Keras session.
Note that this should not be used when training time is paramount, as it disables GPU
computation and CPU parallelism by default for more deterministic computations.
Fix for plotting training history with early stopping callback (thanks to @JamesAllingham).
Return R training history object from
to_numpy_array() function to
keras_array() reflecting automatic use
of Keras default backend float type and "C" ordering.
Add standard layer arguments (e.g.
trainable, etc.) to merge layers
Better support for training models from data tensors in TensorFlow (e.g. Datasets, TFRecords). Add a related example script.
clone_model() function, enabling to construct a new model, given an existing model to use as a template. Works even in a TensorFlow graph different from that of the original model.
target_tensors argument in
compile(), enabling to use custom tensors or placeholders as model targets.
steps_per_epoch argument in
fit(), enabling to train a model from data tensors in a way that is consistent with training from arrays. Similarly, add
steps argument in
layer_subtract() layer function.
weighted_metrics argument in compile to specify metric functions meant to take into account
Enable stateful RNNs with CNTK.
install_keras() function which installs both TensorFlow and Keras
Use keras package as default implementation rather than tf.contrib.keras
Training metrics plotted in realtime within the RStudio Viewer during fit
unserialize_model() functions for saving
Keras models as 'raw' R objects.
Automatically convert 64-bit R floats to backend default float type
Ensure that arrays passed to generator functions are normalized to C-order
to_numpy_array() utility function for custom generators (enables
custom generators to yield C-ordered arrays of the correct float type)
write_grads arguments to
return_state argument to recurrent layers.
tf_config() from tensorflow
is_keras_available() function to probe whether the Keras python
package is available in the current environment.
as.data.frame() S3 method for Keras training history
Remove names from
Return result of
evaluate() as named list
Write run metrics and evaluation data to tfruns
Provide hint to use r-tensorflow environment when importing keras