[Enhance] Use graph transform to deal with more general cases for efficient_conv_bn_eval (#1259)

mmengine/model/efficient_conv_bn_eval.py

@@ -1,5 +1,4 @@
 # Copyright (c) OpenMMLab. All rights reserved.
-from functools import partial
 from operator import attrgetter
 from typing import List, Union
 
@@ -58,48 +57,32 @@ def efficient_conv_bn_eval_forward(bn: nn.modules.batchnorm._BatchNorm,
     return conv._conv_forward(x, weight_on_the_fly, bias_on_the_fly)
 
 
-def bn_once_identity_forward(bn: nn.modules.batchnorm._BatchNorm,
-                             x: torch.Tensor):
-    """The forward function is an identity function.
-
-    The magic is that after one call, the `bn.forward` will be restored to what
-    it used to be.
-    """
-    bn.__dict__.pop('forward')
-    return x
-
-
 def efficient_conv_bn_eval_control(bn: nn.modules.batchnorm._BatchNorm,
                                    conv: nn.modules.conv._ConvNd,
                                    x: torch.Tensor):
     """This function controls whether to use `efficient_conv_bn_eval_forward`.
 
     If the following `bn` is in `eval` mode, then we turn on the special
-    `efficient_conv_bn_eval_forward` and let the following call of `bn.forward`
+    `efficient_conv_bn_eval_forward`.
-    to be identity. Note that this `bn.forward` modification only works for one
-    call. After the call, `bn.forward` will be restored to the default
-    function. This is to deal with the case where one `bn` module is used in
-    multiple places.
     """
     if not bn.training:
         # bn in eval mode
         output = efficient_conv_bn_eval_forward(bn, conv, x)
-        bn.forward = partial(bn_once_identity_forward, bn)
         return output
     else:
-        return conv._conv_forward(x, conv.weight, conv.bias)
+        conv_out = conv._conv_forward(x, conv.weight, conv.bias)
+        return bn(conv_out)
 
 
-def turn_on_efficient_conv_bn_eval_for_single_model(model: torch.nn.Module):
+def efficient_conv_bn_eval_graph_transform(fx_model):
-    # optimize consecutive conv+bn by modifying forward function
+    """Find consecutive conv+bn calls in the graph, inplace modify the graph
-    # Symbolically trace the input model to create an FX GraphModule
+    with the fused operation."""
-    import torch.fx as fx
-    fx_model: fx.GraphModule = fx.symbolic_trace(model)
     modules = dict(fx_model.named_modules())
 
     patterns = [(torch.nn.modules.conv._ConvNd,
                  torch.nn.modules.batchnorm._BatchNorm)]
 
+    pairs = []
     # Iterate through nodes in the graph to find ConvBN blocks
     for node in fx_model.graph.nodes:
         # If our current node isn't calling a Module then we can ignore it.
@@ -116,26 +99,54 @@ def turn_on_efficient_conv_bn_eval_for_single_model(model: torch.nn.Module):
         if not found_pair or len(node.args[0].users) > 1:
             continue
 
-        # check if the conv modules are used in multiple nodes
+        # Find a pair of conv and bn computation nodes to optimize
-        conv_name = node.args[0].target
+        conv_node = node.args[0]
-        bn_name = node.target
+        bn_node = node
+        pairs.append([conv_node, bn_node])
 
-        conv_usage_count = 0
+    for conv_node, bn_node in pairs:
-        for _node in fx_model.graph.nodes:
+        # set insertion point
-            if _node.op != 'call_module':
+        fx_model.graph.inserting_before(conv_node)
-                continue
+        # create `get_attr` node to access modules
-            if _node.target == conv_name:
+        # note that we directly call `create_node` to fill the `name`
-                conv_usage_count += 1
+        # argument. `fx_model.graph.get_attr` and
+        # `fx_model.graph.call_function` does not allow the `name` argument.
+        conv_get_node = fx_model.graph.create_node(
+            op='get_attr', target=conv_node.target, name='get_conv')
+        bn_get_node = fx_model.graph.create_node(
+            op='get_attr', target=bn_node.target, name='get_bn')
+        # prepare args for the fused function
+        args = (bn_get_node, conv_get_node, conv_node.args[0])
+        # create a new node
+        new_node = fx_model.graph.create_node(
+            op='call_function',
+            target=efficient_conv_bn_eval_control,
+            args=args,
+            name='efficient_conv_bn_eval')
+        # this node replaces the original conv + bn, and therefore
+        # should replace the uses of bn_node
+        bn_node.replace_all_uses_with(new_node)
+        # take care of the deletion order:
+        # delete bn_node first, and then conv_node
+        fx_model.graph.erase_node(bn_node)
+        fx_model.graph.erase_node(conv_node)
 
-        if conv_usage_count > 1:
+    # regenerate the code
-            continue
+    fx_model.graph.lint()
+    fx_model.recompile()
 
-        # Find a pair of conv and bn to optimize
-        conv_module = modules[conv_name]
-        bn_module = modules[bn_name]
 
-        conv_module.forward = partial(efficient_conv_bn_eval_control,
+def turn_on_efficient_conv_bn_eval_for_single_model(model: torch.nn.Module):
-                                      bn_module, conv_module)
+    import torch.fx as fx
+
+    # currently we use `fx.symbolic_trace` to trace models.
+    # in the future, we might turn to pytorch 2.0 compile infrastructure to
+    # get the `fx.GraphModule` IR. Nonetheless, the graph transform function
+    # can remain unchanged. We just need to change the way
+    # we get `fx.GraphModule`.
+    fx_model: fx.GraphModule = fx.symbolic_trace(model)
+    efficient_conv_bn_eval_graph_transform(fx_model)
+    model.forward = fx_model.forward
 
 
 def turn_on_efficient_conv_bn_eval(model: torch.nn.Module,

tests/test_model/test_efficient_conv_bn_eval.py

@@ -37,8 +37,8 @@ class BackboneModel(nn.Module):
             x = self.mod1(x)
         # this conv-bn pair can use efficient_conv_bn_eval feature
         x = self.bn1(self.conv1(x))
-        # this conv-bn pair cannot use efficient_conv_bn_eval feature
+        # this conv-bn pair can use efficient_conv_bn_eval feature
-        # because `self.conv2` is used twice
+        # only for the second `self.conv2` call.
         x = self.bn2(self.conv2(self.conv2(x)))
         # this conv-bn pair can use efficient_conv_bn_eval feature
         # just for the first forward of the `self.bn3`