MO fusing activations (#1942)

* Added HSwish operation

* Added HSwish fusing transformation

* Fixed BOM

* Added unit test for HSwish fusing transformation

* Fixed unit tests for transformations using 'build_graph_with_edge_attrs' function to build the graph

* Added fusion transformation for Swish operation

* Added fusing transformation for Softplus operation

* Added fusion transformation for Mish operation

* Added check for the node name in the unit tests

* Fixed Mish fusion pattern

* Updated Mish fusion transformation. Added unit test

* Updated HSwish fusing transformation

* Updated Swish fusion transformation and tests

* Fixed unit tests

model-optimizer/automation/package_BOM.txt

@@ -127,6 +127,7 @@ extensions/front/freeze_placeholder_value.py
 extensions/front/GeLUMerger_Erf.py
 extensions/front/GeLUMerger_Tanh.py
 extensions/front/global_pooling_to_reduce.py
+extensions/front/HSwish_fusion.py
 extensions/front/image_scaler.py
 extensions/front/input_cut.py
 extensions/front/instance_normalization.py
@@ -150,6 +151,7 @@ extensions/front/LayerNorm.py
 extensions/front/Log1p.py
 extensions/front/LogSoftmax.py
 extensions/front/MatMul_normalizer.py
+extensions/front/Mish_fusion.py
 extensions/front/MoveEmbeddedInputsToInputs.py
 extensions/front/mxnet/__init__.py
 extensions/front/mxnet/activation.py
@@ -326,11 +328,13 @@ extensions/front/reshape_dim_normalizer.py
 extensions/front/restore_ports.py
 extensions/front/scatter_normalizer.py
 extensions/front/softmax.py
+extensions/front/Softplus_fusion.py
 extensions/front/softsign_replacer.py
 extensions/front/split_normalizer.py
 extensions/front/SqueezeNormalize.py
 extensions/front/standalone_const_eraser.py
 extensions/front/sub.py
+extensions/front/Swish_fusion.py
 extensions/front/tf/__init__.py
 extensions/front/tf/activation_ext.py
 extensions/front/tf/argmax_ext.py

model-optimizer/extensions/front/HSwish_fusing_test.py

@@ -0,0 +1,196 @@
+"""
+ Copyright (C) 2018-2020 Intel Corporation
+
+ Licensed under the Apache License, Version 2.0 (the "License");
+ you may not use this file except in compliance with the License.
+ You may obtain a copy of the License at
+
+      http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing, software
+ distributed under the License is distributed on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language governing permissions and
+ limitations under the License.
+"""
+
+import unittest
+
+from extensions.front.HSwish_fusion import HSwishWithClamp, HSwishWithMinMax
+from mo.front.common.partial_infer.utils import float_array
+from mo.utils.ir_engine.compare_graphs import compare_graphs
+from mo.utils.unittest.graph import build_graph, const, regular_op, result, build_graph_with_edge_attrs
+
+ref_nodes = {**regular_op('input', {'type': 'Parameter'}),
+             **regular_op('hswish', {'type': 'HSwish', 'name': 'final_mul'}),
+             **result('result')
+             }
+ref_edges = [('input', 'hswish'), ('hswish', 'result')]
+
+
+class HSwishWithClampTest(unittest.TestCase):
+    nodes = {
+        **regular_op('input', {'type': 'Parameter'}),
+        **regular_op('add', {'op': 'Add'}),
+        **regular_op('relu6', {'op': 'Clamp'}),
+        **regular_op('mul', {'op': 'Mul'}),
+        **regular_op('mul_2', {'op': 'Mul', 'name': 'final_mul'}),
+        **const('const_0', float_array([0.0])),
+        **const('const_3', float_array([3.0])),
+        **const('const_6', float_array([6.0])),
+        **const('const_1_6', float_array([1.0 / 6.0])),
+        **result('result'),
+    }
+
+    edges = [('input', 'mul', {'in': 0, 'out': 0}),
+             ('input', 'add', {'in': 0, 'out': 0}),
+             ('const_3', 'add', {'in': 1, 'out': 0}),
+             ('add', 'relu6', {'in': 0, 'out': 0}),
+             ('const_0', 'relu6', {'in': 1, 'out': 0}),
+             ('const_6', 'relu6', {'in': 2, 'out': 0}),
+             ('relu6', 'mul', {'in': 1, 'out': 0}),
+             ('mul', 'mul_2', {'in': 0, 'out': 0}),
+             ('const_1_6', 'mul_2', {'in': 1, 'out': 0}),
+             ('mul_2', 'result', {'in': 0, 'out': 0})]
+
+    def test_hswish_with_clamp(self):
+        graph = build_graph_with_edge_attrs(self.nodes, self.edges, {})
+
+        graph_ref = build_graph(ref_nodes, ref_edges)
+        graph.stage = 'front'
+
+        HSwishWithClamp().find_and_replace_pattern(graph)
+
+        (flag, resp) = compare_graphs(graph, graph_ref, 'result')
+        self.assertTrue(flag, resp)
+        self.assertTrue(len(graph.get_op_nodes(name='final_mul')) == 1 and
+                        graph.get_op_nodes(name='final_mul')[0].op == 'HSwish')
+
+    def test_hswish_with_clamp_wrong_constant(self):
+        graph = build_graph_with_edge_attrs(self.nodes, self.edges, {'const_0': {'value': float_array([0.00001])}})
+
+        graph_ref = graph.copy()
+        graph.stage = 'front'
+
+        HSwishWithClamp().find_and_replace_pattern(graph)
+
+        (flag, resp) = compare_graphs(graph, graph_ref, 'result')
+        self.assertTrue(flag, resp)
+
+    def test_hswish_with_clamp_different_tensors(self):
+        graph = build_graph_with_edge_attrs({
+            **regular_op('input', {'type': 'Parameter'}),
+            **regular_op('input_2', {'type': 'Parameter'}),
+            **regular_op('add', {'op': 'Add'}),
+            **regular_op('relu6', {'op': 'Clamp'}),
+            **regular_op('mul', {'op': 'Mul'}),
+            **regular_op('mul_2', {'op': 'Mul', 'name': 'final_mul'}),
+            **const('const_0', float_array([0.0])),
+            **const('const_3', float_array([3.0])),
+            **const('const_6', float_array([6.0])),
+            **const('const_1_6', float_array([1.0 / 6.0])),
+            **result('result'),
+        }, [('input', 'mul', {'in': 0, 'out': 0}),
+            ('input_2', 'add', {'in': 0, 'out': 0}),
+            ('const_3', 'add', {'in': 1, 'out': 0}),
+            ('add', 'relu6', {'in': 0, 'out': 0}),
+            ('const_0', 'relu6', {'in': 1, 'out': 0}),
+            ('const_6', 'relu6', {'in': 2, 'out': 0}),
+            ('relu6', 'mul', {'in': 1, 'out': 0}),
+            ('mul', 'mul_2', {'in': 0, 'out': 0}),
+            ('const_1_6', 'mul_2', {'in': 1, 'out': 0}),
+            ('mul_2', 'result', {'in': 0, 'out': 0})])
+
+        graph_ref = graph.copy()
+        graph.stage = 'front'
+
+        HSwishWithClamp().find_and_replace_pattern(graph)
+
+        (flag, resp) = compare_graphs(graph, graph_ref, 'result')
+        self.assertTrue(flag, resp)
+
+
+class HSwishWithMinMaxTest(unittest.TestCase):
+    nodes = {
+        **regular_op('input', {'type': 'Parameter'}),
+        **regular_op('add', {'op': 'Add'}),
+        **regular_op('max', {'op': 'Maximum'}),
+        **regular_op('min', {'op': 'Minimum'}),
+        **regular_op('mul', {'op': 'Mul'}),
+        **regular_op('mul_2', {'op': 'Mul', 'name': 'final_mul'}),
+        **const('const_0', float_array([0.0])),
+        **const('const_3', float_array([3.0])),
+        **const('const_6', float_array([6.0])),
+        **const('const_1_6', float_array([1.0 / 6.0])),
+        **result('result'),
+    }
+
+    edges = [('input', 'mul', {'in': 1, 'out': 0}),
+             ('input', 'add', {'in': 0, 'out': 0}),
+             ('const_3', 'add', {'in': 1, 'out': 0}),
+             ('add', 'max', {'in': 0, 'out': 0}),
+             ('const_0', 'max', {'in': 1, 'out': 0}),
+             ('max', 'min', {'in': 0, 'out': 0}),
+             ('const_6', 'min', {'in': 1, 'out': 0}),
+             ('min', 'mul', {'in': 0, 'out': 0}),
+             ('mul', 'mul_2', {'in': 0, 'out': 0}),
+             ('const_1_6', 'mul_2', {'in': 1, 'out': 0}),
+             ('mul_2', 'result', {'in': 0, 'out': 0})]
+
+    def test_hswish_with_min_max(self):
+        graph = build_graph_with_edge_attrs(self.nodes, self.edges, {})
+
+        graph_ref = build_graph(ref_nodes, ref_edges)
+        graph.stage = 'front'
+
+        HSwishWithMinMax().find_and_replace_pattern(graph)
+
+        (flag, resp) = compare_graphs(graph, graph_ref, 'result')
+        self.assertTrue(flag, resp)
+        self.assertTrue(len(graph.get_op_nodes(name='final_mul')) == 1 and
+                        graph.get_op_nodes(name='final_mul')[0].op == 'HSwish')
+
+    def test_hswish_with_min_max_wrong_constant(self):
+        graph = build_graph_with_edge_attrs(self.nodes, self.edges, {'const_0': {'value': float_array([0.00001])}})
+
+        graph_ref = graph.copy()
+        graph.stage = 'front'
+
+        HSwishWithMinMax().find_and_replace_pattern(graph)
+
+        (flag, resp) = compare_graphs(graph, graph_ref, 'result')
+        self.assertTrue(flag, resp)
+
+    def test_hswish_with_min_max_different_tensors(self):
+        graph = build_graph_with_edge_attrs({
+            **regular_op('input', {'type': 'Parameter'}),
+            **regular_op('input_2', {'type': 'Parameter'}),
+            **regular_op('add', {'op': 'Add'}),
+            **regular_op('max', {'op': 'Maximum'}),
+            **regular_op('min', {'op': 'Minimum'}),
+            **regular_op('mul', {'op': 'Mul'}),
+            **regular_op('mul_2', {'op': 'Mul', 'name': 'final_mul'}),
+            **const('const_0', float_array([0.0])),
+            **const('const_3', float_array([3.0])),
+            **const('const_6', float_array([6.0])),
+            **const('const_1_6', float_array([1.0 / 6.0])),
+            **result('result'),
+        }, [('input_2', 'mul', {'in': 1, 'out': 0}),
+            ('input', 'add', {'in': 0, 'out': 0}),
+            ('const_3', 'add', {'in': 1, 'out': 0}),
+            ('add', 'max', {'in': 0, 'out': 0}),
+            ('const_0', 'max', {'in': 1, 'out': 0}),
+            ('max', 'min', {'in': 0, 'out': 0}),
+            ('const_6', 'min', {'in': 1, 'out': 0}),
+            ('min', 'mul', {'in': 0, 'out': 0}),
+            ('mul', 'mul_2', {'in': 0, 'out': 0}),
+            ('const_1_6', 'mul_2', {'in': 1, 'out': 0}),
+            ('mul_2', 'result', {'in': 0, 'out': 0})])
+
+        graph_ref = graph.copy()
+        graph.stage = 'front'
+
+        HSwishWithMinMax().find_and_replace_pattern(graph)
+
+        (flag, resp) = compare_graphs(graph, graph_ref, 'result')
+        self.assertTrue(flag, resp)

model-optimizer/extensions/front/HSwish_fusion.py

@@ -0,0 +1,123 @@
+"""
+ Copyright (C) 2018-2020 Intel Corporation
+
+ Licensed under the Apache License, Version 2.0 (the "License");
+ you may not use this file except in compliance with the License.
+ You may obtain a copy of the License at
+
+      http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing, software
+ distributed under the License is distributed on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language governing permissions and
+ limitations under the License.
+"""
+import numpy as np
+
+from extensions.front.AttributedClampNormalizer import AttributedClampNormalizer
+from extensions.ops.activation_ops import HSwish
+from mo.front.common.replacement import FrontReplacementSubgraph
+from mo.front.subgraph_matcher import SubgraphMatch
+from mo.graph.graph import Graph, rename_nodes
+
+
+def replace_with_hswish(graph: Graph, match: [dict, SubgraphMatch]):
+    add = match['add']
+    mul = match['mul']
+    mul_2 = match['mul_2']
+
+    # determine the input port of Add and Mul which gets the 'input' node output
+    add_input_port_idx = int(add.in_port(0).get_connection().get_source().node.soft_get('op') == 'Const')
+    mul_input_port_idx = int(mul.in_port(0).get_connection().get_source().node.soft_get('op') in ['Clamp', 'Minimum'])
+
+    # check that the same tensor provided as input to Add and Mul
+    if add.in_port(add_input_port_idx).get_source() != mul.in_port(mul_input_port_idx).get_source():
+        return
+    mul_2_name = mul_2.soft_get('name', mul_2.id)
+
+    hswish = HSwish(graph, {}).create_node()
+    hswish.in_port(0).connect(add.in_port(add_input_port_idx).get_source())
+    mul_2.out_port(0).get_connection().set_source(hswish.out_port(0))
+
+    rename_nodes([(mul_2, mul_2_name + '/TBR'), (hswish, mul_2_name)])
+
+
+class HSwishWithClamp(FrontReplacementSubgraph):
+    """
+    The transformation looks for the pattern with ReLU6 (Clamp) defining the HSwish function:
+    HSwish(x) = x * Relu6(x + 3) / 6.0.
+    """
+    enabled = True
+
+    def run_after(self):
+        return [AttributedClampNormalizer]
+
+    def pattern(self):
+        return dict(
+            nodes=[
+                ('input', dict()),
+                ('add', dict(op='Add')),
+                ('const_0', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 0.0, atol=1e-6))),
+                ('const_3', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 3.0, atol=1e-6))),
+                ('const_6', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 6.0, atol=1e-6))),
+                ('const_1_6', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 1 / 6.0, atol=1e-6))),
+                ('clamp', dict(op='Clamp')),
+                ('mul', dict(op='Mul')),
+                ('mul_2', dict(op='Mul')),
+            ],
+            edges=[
+                ('input', 'add', {}),
+                ('input', 'mul', {}),
+                ('const_3', 'add', {}),
+                ('add', 'clamp', {'in': 0}),
+                ('const_0', 'clamp', {'in': 1}),
+                ('const_6', 'clamp', {'in': 2}),
+                ('clamp', 'mul', {}),
+                ('mul', 'mul_2', {}),
+                ('const_1_6', 'mul_2', {}),
+            ])
+
+    def replace_sub_graph(self, graph: Graph, match: [dict, SubgraphMatch]):
+        replace_with_hswish(graph, match)
+
+
+class HSwishWithMinMax(FrontReplacementSubgraph):
+    """
+    The transformation looks for the pattern with Min/Max defining the HSwish function:
+    HSwish(x) = x * Min(Max(x + 3, 0), 6) / 6.0.
+    """
+    enabled = True
+
+    def run_after(self):
+        return [AttributedClampNormalizer]
+
+    def pattern(self):
+        return dict(
+            nodes=[
+                ('input', dict()),
+                ('add', dict(op='Add')),
+                ('const_0', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 0.0, atol=1e-6))),
+                ('const_3', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 3.0, atol=1e-6))),
+                ('const_6', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 6.0, atol=1e-6))),
+                ('const_1_6', dict(op='Const', value=lambda v: v is not None and np.isclose(v, 1 / 6.0, atol=1e-6))),
+                ('max', dict(op='Maximum')),
+                ('min', dict(op='Minimum')),
+                ('mul', dict(op='Mul')),
+                ('mul_2', dict(op='Mul')),
+            ],
+            edges=[
+                ('input', 'add', {'out': 0}),
+                ('input', 'mul', {'out': 0}),
+                ('const_3', 'add', {}),
+                ('add', 'max', {}),
+                ('const_0', 'max', {}),
+                ('max', 'min', {}),
+                ('const_6', 'min', {}),
+                ('min', 'mul', {}),
+                ('mul', 'mul_2', {}),
+                ('const_1_6', 'mul_2', {}),
+            ])
+
+    def replace_sub_graph(self, graph: Graph, match: [dict, SubgraphMatch]):
+        replace_with_hswish(graph, match)