Improve parsing of functional layers arguments. Support nn.functional…

….layer_norm and substract

model_compression_toolkit/core/pytorch/constants.py

@@ -42,7 +42,6 @@
 OP_CALL_ARGS = 'op_call_args'
 OP_CALL_KWARGS = 'op_call_kwargs'
 INPUTS_AS_LIST = 'inputs_as_list'
-OP_CALL_ARGS_ORDER = 'op_call_args_order'
 INPLACE = 'inplace'
 HARDTANH_MIN_VAL = 'min_val'
 HARDTANH_MAX_VAL = 'max_val'

model_compression_toolkit/core/pytorch/reader/graph_builders.py

@@ -264,4 +264,4 @@ def edges_builder(model: GraphModule,
                 edges.append(
                     Edge(fx_node_2_graph_node[node], fx_node_2_graph_node[out_node], src_index, dst_index))
 
-    return edges
+    return edges

tests/pytorch_tests/model_tests/test_feature_models_runner.py

@@ -20,7 +20,6 @@
 import model_compression_toolkit as mct
 from model_compression_toolkit.gptq.common.gptq_config import RoundingType
 from tests.pytorch_tests.model_tests.feature_models.add_net_test import AddNetTest
-from tests.pytorch_tests.model_tests.feature_models.sub_net_test import SubNetTest
 from tests.pytorch_tests.model_tests.feature_models.layer_norm_net_test import LayerNormNetTest
 from tests.pytorch_tests.model_tests.feature_models.conv2d_replacement_test import DwConv2dReplacementTest
 from tests.pytorch_tests.model_tests.feature_models.mixed_precision_bops_test import MixedPrecisionBopsBasicTest, \
@@ -123,13 +122,6 @@ def test_add_net(self):
         """
         AddNetTest(self).run_test()
 
-    def test_sub_net(self):
-        """
-        This tests check the subtraction operations.
-        Both with different layers and with constants.
-        """
-        SubNetTest(self).run_test()
-
     def test_layer_norm_net(self):
         """
         This tests check the nn.functional.layer_norm operations.