diff --git a/onnx2pytorch/convert/operations.py b/onnx2pytorch/convert/operations.py
index abdaea3..bdd6e29 100644
--- a/onnx2pytorch/convert/operations.py
+++ b/onnx2pytorch/convert/operations.py
@@ -75,6 +75,15 @@ def convert_operations(onnx_graph, opset_version, batch_dim=0, enable_pruning=Tr
     iterator: (op_id, op_name, op)
     """
     weights = {tensor.name: tensor for tensor in onnx_graph.initializer}
+    branching_factor = {}
+
+    # `len(node.output)` is unreliable, so we need to count ourselves
+    for i, node in enumerate(onnx_graph.node):
+        if len(node.input) > 0:
+            input_id = node.input[0]
+            if input_id not in branching_factor:
+                branching_factor[input_id] = 0
+            branching_factor[input_id] += 1
 
     for i, node in enumerate(onnx_graph.node):
         # extract only useful inputs
@@ -107,7 +116,8 @@ def convert_operations(onnx_graph, opset_version, batch_dim=0, enable_pruning=Tr
         elif node.op_type == "Div":
             op = Div()
         elif node.op_type == "Elu":
-            op = nn.ELU(**extract_attributes(node), inplace=True)
+            is_input_branching = branching_factor[node.input[0]] > 1
+            op = nn.ELU(**extract_attributes(node), inplace=not is_input_branching)
         elif node.op_type == "Equal":
             op = OperatorWrapper(torch.eq)
         elif node.op_type == "Erf":
@@ -136,7 +146,10 @@ def convert_operations(onnx_graph, opset_version, batch_dim=0, enable_pruning=Tr
         elif node.op_type == "InstanceNormalization":
             op = convert_instance_norm_layer(node, params=params)
         elif node.op_type == "LeakyRelu":
-            op = nn.LeakyReLU(**extract_attributes(node), inplace=True)
+            is_input_branching = branching_factor[node.input[0]] > 1
+            op = nn.LeakyReLU(
+                **extract_attributes(node), inplace=not is_input_branching
+            )
         elif node.op_type == "Less":
             op = OperatorWrapper(torch.less)
         elif node.op_type == "Log":
@@ -215,7 +228,8 @@ def convert_operations(onnx_graph, opset_version, batch_dim=0, enable_pruning=Tr
         elif node.op_type == "ReduceSum":
             op = ReduceSum(opset_version=opset_version, **extract_attributes(node))
         elif node.op_type == "Relu":
-            op = nn.ReLU(inplace=True)
+            is_input_branching = branching_factor[node.input[0]] > 1
+            op = nn.ReLU(inplace=not is_input_branching)
         elif node.op_type == "Reshape":
             shape = list(
                 filter(lambda x: x.name == node.input[1], onnx_graph.initializer)