EagerSession.DevicePlacementPolicy

public static final enum EagerSession.DevicePlacementPolicy

Controls how to act when we try to run an operation on a given device but some input tensors are not on that device.

final int
compareTo(E arg0)
int
compareTo(Object arg0)
final boolean
equals(Object arg0)
final Class<E>
getDeclaringClass()
final int
hashCode()
final String
name()
final int
ordinal()
String
toString()
static <T extends Enum<T>> T
valueOf(Class<T> arg0, String arg1)
boolean
equals(Object arg0)
final Class<?>
getClass()
int
hashCode()
final void
notify()
final void
notifyAll()
String
toString()
final void
wait(long arg0, int arg1)
final void
wait(long arg0)
final void
wait()
abstract int
compareTo(E extends Enum<E> arg0)

Running operations with input tensors on the wrong device will fail.

public static final EagerSession.DevicePlacementPolicy SILENT

Silently copy the tensor, which has a performance cost since the operation will be blocked till the copy completes. This is the default placement policy.

public static final EagerSession.DevicePlacementPolicy SILENT_FOR_INT32

Placement policy which silently copies int32 tensors but not other dtypes.

public static final EagerSession.DevicePlacementPolicy WARN

Copy the tensor to the right device but log a warning.