def test_infer_constant_input_sum(self):
graph = build_graph(nodes_attributes2, edges2, inputs2)
sparse_segment_sum_node = Node(graph, 'sparse_segment_sum_node')
SparseSegmentSum.infer(sparse_segment_sum_node)
# prepare reference results
ref_output_segments_shape = int64_array([2, 4])
ref_output_segments_value = np.array([[0, 0, 0, 0], [5, 6, 7, 8]],
dtype=np.float)
# get resulted shapes
res_output_segments_shape = graph.node['output_segments']['shape']
res_output_segments_value = graph.node['output_segments']['value']
self.assertTrue(
np.array_equal(ref_output_segments_shape,
res_output_segments_shape),
'Shapes do not match expected: {} and given: {}'.format(
ref_output_segments_shape, res_output_segments_shape))
self.assertTrue(
np.array_equal(ref_output_segments_value,
res_output_segments_value),
'Shapes do not match expected: {} and given: {}'.format(
ref_output_segments_value, res_output_segments_value))
def test_partial_infer(self):
graph = build_graph(nodes_attributes1, edges1, inputs1)
sparse_segment_sum_node = Node(graph, 'sparse_segment_sum_node')
SparseSegmentSum.infer(sparse_segment_sum_node)
# prepare reference results
ref_output_segments_shape = int64_array([40, 4, 5])
# get resulted shapes
res_output_segments_shape = graph.node['output_segments']['shape']
self.assertTrue(np.array_equal(ref_output_segments_shape, res_output_segments_shape),
'Shapes do not match expected: {} and given: {}'.format(ref_output_segments_shape, res_output_segments_shape))
def extract(cls, node):
attrs = {}
SparseSegmentSum.update_node_stat(node, attrs)
return cls.enabled
def extract(node):
attrs = {}
SparseSegmentSum.update_node_stat(node, attrs)
return __class__.enabled