Type tf.keras.mixed_precision.experimental
Namespace tensorflow
Methods
Properties
Public static methods
object global_policy()
Returns the global Policy. The global policy is the default policy used for layers, if no policy is
passed to the layer constructor. If no policy has been set with
`keras.mixed_precision.experimental.set_policy`, this will return a policy
constructed from `tf.keras.backend.floatx()` in TensorFlow 2, or an "infer"
policy in TensorFlow 1. See `keras.mixed_precision.experimental.Policy` for more information.
Returns
-
object - The global Policy.
object global_policy_dyn()
Returns the global Policy. The global policy is the default policy used for layers, if no policy is
passed to the layer constructor. If no policy has been set with
`keras.mixed_precision.experimental.set_policy`, this will return a policy
constructed from `tf.keras.backend.floatx()` in TensorFlow 2, or an "infer"
policy in TensorFlow 1. See `keras.mixed_precision.experimental.Policy` for more information.
Returns
-
object - The global Policy.
void set_policy(Policy policy)
Sets the global Policy. The global policy is the default policy used for layers, if no policy is
passed to the layer constructor. If no global policy is set, layers will
instead default to a Policy constructed from `tf.keras.backend.floatx()` in
TensorFlow 2. In TensorFlow 1, layers default to an "infer" policy. See `keras.mixed_precision.experimental.Policy` for more information.
Parameters
-
Policypolicy - A Policy, or a string that will be converted to a Policy..
void set_policy(string policy)
Sets the global Policy. The global policy is the default policy used for layers, if no policy is
passed to the layer constructor. If no global policy is set, layers will
instead default to a Policy constructed from `tf.keras.backend.floatx()` in
TensorFlow 2. In TensorFlow 1, layers default to an "infer" policy. See `keras.mixed_precision.experimental.Policy` for more information.
Parameters
-
stringpolicy - A Policy, or a string that will be converted to a Policy..
object set_policy_dyn(object policy)
Sets the global Policy. The global policy is the default policy used for layers, if no policy is
passed to the layer constructor. If no global policy is set, layers will
instead default to a Policy constructed from `tf.keras.backend.floatx()` in
TensorFlow 2. In TensorFlow 1, layers default to an "infer" policy. See `keras.mixed_precision.experimental.Policy` for more information.
Parameters
-
objectpolicy - A Policy, or a string that will be converted to a Policy..