def testOpDefDefaultType(self):
im = np.random.randint( # pylint: disable=too-many-function-args
low=0,
high=65535,
size=100,
dtype=np.uint16).reshape(10, 10, 1)
context.ensure_initialized()
fastpath_dtype = test_ops.dtype_with_default_op(im).numpy()
slowpath_dtype = test_ops.dtype_with_default_op_eager_fallback(
im, None, context.context()).numpy()
# Ensure the fastpath and slowpath eager paths work.
self.assertEqual(fastpath_dtype, slowpath_dtype)
with ops.Graph().as_default(), self.cached_session():
graph_dtype_symbolic = test_ops.dtype_with_default_op(im)
graph_dtype = self.evaluate(graph_dtype_symbolic)
# Ensure the eager path matches the graph path.
self.assertEqual(fastpath_dtype, graph_dtype)
# Unfortunately, as of now, this doesn't work as expected on def_functions,
# since we convert the numpy arrays to tensors pre-tracing (which won't get
# overriddent by the default type).
@def_function.function
def func(im):
return test_ops.dtype_with_default_op(im)
function_dtype = func(im).numpy()
self.assertNotEqual(fastpath_dtype, function_dtype)
# Captures are OK, since they don't go through the conversion path.
@def_function.function
def func_captured():
return test_ops.dtype_with_default_op(im)
function_dtype = func_captured().numpy()
self.assertEqual(fastpath_dtype, function_dtype)
def func_captured():
return test_ops.dtype_with_default_op(im)
def func(im):
return test_ops.dtype_with_default_op(im)
