tensorflow 2.14.0
-
install_tensorflow()
changes:- Installs TensorFlow v2.14 by default.
- Now will automatically install the required Nvidia CUDA runtime as a pip
package if on Linux and a GPU is detected. You can opt-out by passing
install_tensorflow(cuda = FALSE)
. Aside from the Nvidia driver, no other
pre-existing Nvidia CUDA packages are now necessary. - The
configure_cudnn
argument is now superseded by the new argumentcuda
. - New argument
metal
, for specifying if thetensorflow-metal
pip package
should be installed on Arm Macs. Defaults toTRUE
on Arm Macs.
-
Fixed an issue where
as.array()
and other methods might fail if the tensor
had conversion disabled viar_to_py()
orconvert = FALSE
. -
Fixed an issue where Ops group generic dispatch would error one object was a tensor
and the other was a non-tensor Python object (e.g., a numpy array). -
Removed long deprecated symbols:
install_tensorflow_extras()
,tfe_enable_eager_execution()
-
tfestimator generics
train()
andtrain_and_evaluate()
now warn about
their deprecation status when called. The will be removed in a future release.