def test_build_all_signature_defs_with_single_alternatives(self):
# Force the test to run in graph mode.
# This tests a deprecated v1 API that depends on graph-only functions such
# as build_tensor_info.
with ops.Graph().as_default():
receiver_tensor = array_ops.placeholder(dtypes.string)
receiver_tensors_alternative_1 = array_ops.placeholder(
dtypes.int64)
receiver_tensors_alternative_2 = array_ops.sparse_placeholder(
dtypes.float32)
# Note we are passing single Tensors as values of
# receiver_tensors_alternatives, where normally that is a dict.
# In this case a dict will be created using the default receiver tensor
# name "input".
receiver_tensors_alternatives = {
"other1": receiver_tensors_alternative_1,
"other2": receiver_tensors_alternative_2
}
output_1 = constant_op.constant([1.])
output_2 = constant_op.constant(["2"])
output_3 = constant_op.constant(["3"])
export_outputs = {
signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
export_output.RegressionOutput(value=output_1),
"head-2":
export_output.ClassificationOutput(classes=output_2),
"head-3":
export_output.PredictOutput(
outputs={"some_output_3": output_3}),
}
signature_defs = export_utils.build_all_signature_defs(
receiver_tensor, export_outputs, receiver_tensors_alternatives)
expected_signature_defs = {
"serving_default":
signature_def_utils.regression_signature_def(
receiver_tensor, output_1),
"head-2":
signature_def_utils.classification_signature_def(
receiver_tensor, output_2, None),
"head-3":
signature_def_utils.predict_signature_def(
{"input": receiver_tensor}, {"some_output_3": output_3}),
"other1:head-3":
signature_def_utils.predict_signature_def(
{"input": receiver_tensors_alternative_1},
{"some_output_3": output_3}),
"other2:head-3":
signature_def_utils.predict_signature_def(
{"input": receiver_tensors_alternative_2},
{"some_output_3": output_3})
# Note that the alternatives 'other:serving_default' and
# 'other:head-2' are invalid, because regression and classification
# signatures must take a single string input. Here we verify that
# these invalid signatures are not included in the export_utils.
}
self.assertDictEqual(expected_signature_defs, signature_defs)
def test_build_all_signature_defs_without_receiver_alternatives(self):
receiver_tensor = array_ops.placeholder(dtypes.string)
output_1 = constant_op.constant([1.])
output_2 = constant_op.constant(["2"])
output_3 = constant_op.constant(["3"])
export_outputs = {
signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
export_output.RegressionOutput(value=output_1),
"head-2": export_output.ClassificationOutput(classes=output_2),
"head-3": export_output.PredictOutput(outputs={
"some_output_3": output_3
}),
}
signature_defs = export_utils.build_all_signature_defs(
receiver_tensor, export_outputs)
expected_signature_defs = {
"serving_default":
signature_def_utils.regression_signature_def(receiver_tensor,
output_1),
"head-2":
signature_def_utils.classification_signature_def(receiver_tensor,
output_2, None),
"head-3":
signature_def_utils.predict_signature_def({
"input": receiver_tensor
}, {"some_output_3": output_3})
}
self.assertDictEqual(expected_signature_defs, signature_defs)
def test_build_all_signature_defs_without_receiver_alternatives(self):
# Force the test to run in graph mode.
# This tests a deprecated v1 API that depends on graph-only functions such
# as build_tensor_info.
with ops.Graph().as_default():
receiver_tensor = array_ops.placeholder(dtypes.string)
output_1 = constant_op.constant([1.])
output_2 = constant_op.constant(["2"])
output_3 = constant_op.constant(["3"])
export_outputs = {
signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
export_output.RegressionOutput(value=output_1),
"head-2":
export_output.ClassificationOutput(classes=output_2),
"head-3":
export_output.PredictOutput(
outputs={"some_output_3": output_3}),
}
signature_defs = export_utils.build_all_signature_defs(
receiver_tensor, export_outputs)
expected_signature_defs = {
"serving_default":
signature_def_utils.regression_signature_def(
receiver_tensor, output_1),
"head-2":
signature_def_utils.classification_signature_def(
receiver_tensor, output_2, None),
"head-3":
signature_def_utils.predict_signature_def(
{"input": receiver_tensor}, {"some_output_3": output_3})
}
self.assertDictEqual(expected_signature_defs, signature_defs)
def test_build_standardized_signature_def_classify_classes_only(self):
"""Tests classification with one output tensor."""
with context.graph_mode():
input_tensors = {
'input-1':
array_ops.placeholder(
dtypes.string, 1, name='input-tensor-1')
}
classes = array_ops.placeholder(dtypes.string, 1, name='output-tensor-1')
export_output = export_output_lib.ClassificationOutput(classes=classes)
actual_signature_def = export_output.as_signature_def(input_tensors)
expected_signature_def = meta_graph_pb2.SignatureDef()
shape = tensor_shape_pb2.TensorShapeProto(
dim=[tensor_shape_pb2.TensorShapeProto.Dim(size=1)])
dtype_string = types_pb2.DataType.Value('DT_STRING')
expected_signature_def.inputs[
signature_constants.CLASSIFY_INPUTS].CopyFrom(
meta_graph_pb2.TensorInfo(name='input-tensor-1:0',
dtype=dtype_string,
tensor_shape=shape))
expected_signature_def.outputs[
signature_constants.CLASSIFY_OUTPUT_CLASSES].CopyFrom(
meta_graph_pb2.TensorInfo(name='output-tensor-1:0',
dtype=dtype_string,
tensor_shape=shape))
expected_signature_def.method_name = (
signature_constants.CLASSIFY_METHOD_NAME)
self.assertEqual(actual_signature_def, expected_signature_def)
def test_classify_scores_must_be_float(self):
with context.graph_mode():
scores = array_ops.placeholder(dtypes.string,
1,
name='output-tensor-1')
with self.assertRaisesRegexp(
ValueError,
'Classification scores must be a float32 Tensor'):
export_output_lib.ClassificationOutput(scores=scores)
def test_classify_classes_must_be_strings(self):
with context.graph_mode():
classes = array_ops.placeholder(dtypes.float32,
1,
name='output-tensor-1')
with self.assertRaisesRegex(
ValueError,
'Classification classes must be a string Tensor'):
export_output_lib.ClassificationOutput(classes=classes)
def test_build_all_signature_defs_with_dict_alternatives(self):
with context.graph_mode():
receiver_tensor = array_ops.placeholder(dtypes.string)
receiver_tensors_alternative_1 = {
"foo": array_ops.placeholder(dtypes.int64),
"bar": array_ops.sparse_placeholder(dtypes.float32)}
receiver_tensors_alternatives = {"other": receiver_tensors_alternative_1}
output_1 = constant_op.constant([1.])
output_2 = constant_op.constant(["2"])
output_3 = constant_op.constant(["3"])
export_outputs = {
signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
export_output.RegressionOutput(value=output_1),
"head-2": export_output.ClassificationOutput(classes=output_2),
"head-3": export_output.PredictOutput(outputs={
"some_output_3": output_3
}),
}
signature_defs = export_utils.build_all_signature_defs(
receiver_tensor, export_outputs, receiver_tensors_alternatives)
expected_signature_defs = {
"serving_default":
signature_def_utils.regression_signature_def(
receiver_tensor,
output_1),
"head-2":
signature_def_utils.classification_signature_def(
receiver_tensor,
output_2, None),
"head-3":
signature_def_utils.predict_signature_def(
{"input": receiver_tensor},
{"some_output_3": output_3}),
"other:head-3":
signature_def_utils.predict_signature_def(
receiver_tensors_alternative_1,
{"some_output_3": output_3})
# Note that the alternatives 'other:serving_default' and
# 'other:head-2' are invalid, because regession and classification
# signatures must take a single string input. Here we verify that
# these invalid signatures are not included in the export_utils.
}
self.assertDictEqual(expected_signature_defs, signature_defs)
def test_export_outputs_for_mode(self):
predictions = {"predictions": constant_op.constant([1.])}
loss = {"loss": constant_op.constant([2.])}
metrics = {
"metrics": (constant_op.constant([3.]), constant_op.constant([4.]))
}
expected_metrics = {
"metrics/value": metrics["metrics"][0],
"metrics/update_op": metrics["metrics"][1]
}
def _build_export_output(mode):
return export_utils.export_outputs_for_mode(
mode, None, predictions, loss, metrics)
ret = _build_export_output(KerasModeKeys.TRAIN)
self.assertIn(signature_constants.DEFAULT_TRAIN_SIGNATURE_DEF_KEY, ret)
export_out = ret[signature_constants.DEFAULT_TRAIN_SIGNATURE_DEF_KEY]
self.assertIsInstance(export_out, export_output.TrainOutput)
self.assertEqual(export_out.predictions, predictions)
self.assertEqual(export_out.loss, loss)
self.assertEqual(export_out.metrics, expected_metrics)
ret = _build_export_output(KerasModeKeys.TEST)
self.assertIn(signature_constants.DEFAULT_EVAL_SIGNATURE_DEF_KEY, ret)
export_out = ret[signature_constants.DEFAULT_EVAL_SIGNATURE_DEF_KEY]
self.assertIsInstance(export_out, export_output.EvalOutput)
self.assertEqual(export_out.predictions, predictions)
self.assertEqual(export_out.loss, loss)
self.assertEqual(export_out.metrics, expected_metrics)
ret = _build_export_output(KerasModeKeys.PREDICT)
self.assertIn(signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY,
ret)
export_out = ret[signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY]
self.assertIsInstance(export_out, export_output.PredictOutput)
self.assertEqual(export_out.outputs, predictions)
classes = constant_op.constant(["class5"])
ret = export_utils.export_outputs_for_mode(
KerasModeKeys.PREDICT,
{"classify": export_output.ClassificationOutput(classes=classes)})
self.assertIn("classify", ret)
export_out = ret["classify"]
self.assertIsInstance(export_out, export_output.ClassificationOutput)
self.assertEqual(export_out.classes, classes)
def test_classify_requires_classes_or_scores(self):
with self.assertRaisesRegexp(
ValueError, 'At least one of scores and classes must be set.'):
export_output_lib.ClassificationOutput()
def test_classify_requires_classes_or_scores(self):
with self.assertRaisesRegex(
ValueError,
'Cannot create a ClassificationOutput with empty arguments'):
export_output_lib.ClassificationOutput()
def test_build_all_signature_defs_with_dict_alternatives(self):
# Force the test to run in graph mode.
# This tests a deprecated v1 API that depends on graph-only functions such
# as build_tensor_info.
with ops.Graph().as_default():
receiver_tensor = array_ops.placeholder(dtypes.string)
receiver_tensors_alternative_1 = {
"foo": array_ops.placeholder(dtypes.int64),
"bar": array_ops.sparse_placeholder(dtypes.float32)
}
unfed_input = array_ops.placeholder(dtypes.bool)
receiver_tensors_alternative_2 = {"unfed": unfed_input}
receiver_tensors_alternatives = {
"other": receiver_tensors_alternative_1,
"with_unfed_input": receiver_tensors_alternative_2
}
output_1 = constant_op.constant([1.])
output_2 = constant_op.constant(["2"])
output_3 = constant_op.constant(["3"])
output_4 = unfed_input
output_5 = math_ops.logical_not(unfed_input)
export_outputs = {
signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
export_output.RegressionOutput(value=output_1),
"head-2":
export_output.ClassificationOutput(classes=output_2),
"head-3":
export_output.PredictOutput(outputs={"some_output_3": output_3}),
"head-4":
export_output.PredictOutput(outputs={"some_output_4": output_4}),
"head-5":
export_output.PredictOutput(outputs={"some_output_5": output_5}),
}
signature_defs = export_utils.build_all_signature_defs(
receiver_tensor, export_outputs, receiver_tensors_alternatives)
expected_signature_defs = {
"serving_default":
signature_def_utils.regression_signature_def(
receiver_tensor, output_1),
"head-2":
signature_def_utils.classification_signature_def(
receiver_tensor, output_2, None),
"head-3":
signature_def_utils.predict_signature_def(
{"input": receiver_tensor}, {"some_output_3": output_3}),
"other:head-3":
signature_def_utils.predict_signature_def(
receiver_tensors_alternative_1, {"some_output_3": output_3}),
# Note that the alternatives 'other:serving_default' and
# 'other:head-2' are invalid, because regression and classification
# signatures must take a single string input. Here we verify that
# these invalid signatures are not included in the export_utils.
# Similarly, we verify that 'head-4' and 'head-5', which depend on an
# input that is not being fed as a receiver tensor, are also omitted.
# All the three heads are present when that input is fed, however:
"with_unfed_input:head-3":
signature_def_utils.predict_signature_def(
receiver_tensors_alternative_2, {"some_output_3": output_3}),
"with_unfed_input:head-4":
signature_def_utils.predict_signature_def(
receiver_tensors_alternative_2, {"some_output_4": output_4}),
"with_unfed_input:head-5":
signature_def_utils.predict_signature_def(
receiver_tensors_alternative_2, {"some_output_5": output_5})
}
self.assertDictEqual(expected_signature_defs, signature_defs)
