[hotfix] fix some bugs during gpt2 testing (#1379)

colossalai/fx/passes/passes_for_gpt2_test.py

@@ -72,25 +72,53 @@ def split_with_split_nodes_pass_for_gp2_test(annotated_gm: torch.fx.GraphModule)
         new_placeholder_list = []
         for node in gm.graph.nodes:
             if node.op == 'output':
-                output_type = node.args[0].__class__
+                if isinstance(node.args[0], (tuple, list)):
-                output_args.extend(list(node.args[0]))
+                    output_type = node.args[0].__class__
-                for n in node.args[0]:
+                    output_args.extend([n.name for n in node.args[0]])
-                    if next_partition_placeholders and n not in next_partition_placeholders:
+                else:
-                        output_args.remove(n)
+                    output_args.append(node.args[0].name)
+                rm_list = []
+                for name in output_args:
+                    if next_partition_placeholders and name not in next_partition_placeholders:
+                        rm_list.append(name)
+                for name in rm_list:
+                    output_args.remove(name)
                 gm.graph.erase_node(node)
             else:
                 non_output_list.append(node.name)
-        for node in next_partition_placeholders:
+
-            if node not in output_args:
+        for name in next_partition_placeholders:
-                output_args.append(node)
+            if name not in output_args:
-        for node in output_args:
+                output_args.append(name)
-            if node.name not in non_output_list:
+
-                gm.graph.placeholder(node.name)
+        for name in output_args:
+            if name not in non_output_list:
+                gm.graph.placeholder(name)
+
+        # convert name to node for output_args
+        for index, name in enumerate(output_args):
+            for n in gm.graph.nodes:
+                if n.name == name:
+                    output_args[index] = n
+                    continue
+
+        # reorder the output args to make sure
+        # output args has same order as next partition placeholder
+        reorder_output_args = []
+        if next_partition_placeholders:
+            for name in next_partition_placeholders:
+                for node in output_args:
+                    if node.name == name:
+                        reorder_output_args.append(node)
+                        continue
 
         for node in gm.graph.nodes:
             if node.op == 'placeholder':
-                new_placeholder_list.append(node)
+                new_placeholder_list.append(node.name)
-        gm.graph.output(output_type(output_args))
+        if output_type is not None:
+            gm.graph.output(output_type(output_args))
+        else:
+            gm.graph.output(output_args)
         gm.recompile()
         return gm, new_placeholder_list
 
@@ -115,15 +143,6 @@ def split_with_split_nodes_pass_for_gp2_test(annotated_gm: torch.fx.GraphModule)
     for submodule in submodules:
         submodule = eliminate_unused_placeholders(submodule)
         placeholder_dict[submodule] = []
-        for node in submodule.graph.nodes:
-            if node.op == 'placeholder':
-                placeholder_dict[submodule].append(node)
-    output_dict = {}
-    for submodule in submodules:
-        output_dict[submodule] = []
-        for node in submodule.graph.nodes:
-            if node.op == 'output':
-                output_dict[submodule].append(node.name)
     submodules.reverse()
     for index, submodule in enumerate(submodules):
         if index == 0:
@@ -297,7 +316,7 @@ def split_module_for_gpt2_test(
                                                    name=node.name)
             new_node.meta = node.meta.copy()
             partition.environment[node] = new_node
-    assert 'add_85' in orig_nodes
+
     # Set up values to construct base module
     base_mod_env: Dict[str, torch.fx.node.Node] = {}
     base_mod_graph: torch.fx.graph.Graph = torch.fx.graph.Graph()