Add backward function hook to dygraph (#37141)

paddle/fluid/imperative/basic_engine.cc

@@ -569,6 +569,13 @@ void BasicEngine::Execute() {
         }
       }
 
+      // Function Post Hook
+      if (cur_op.HasVoidFunctionPostHook()) {
+        for (const auto& hook : cur_op.GetVoidFunctionPostHooks()) {
+          (*hook)();
+        }
+      }
+
       for (auto& pair : inplace_output_grad_var_list_) {
         *pair.first = std::move(*pair.second);
       }

paddle/fluid/imperative/op_base.h

@@ -186,6 +186,19 @@ class OpBase {
 
   static pten::KernelContext* GetKernelContext() { return &pt_kernel_context_; }
 
+  bool HasVoidFunctionPostHook() const {
+    return !void_function_post_hooks_.empty();
+  }
+
+  void AddVoidFunctionPostHook(std::shared_ptr<std::function<void()>>&& hook) {
+    void_function_post_hooks_.emplace_back(std::move(hook));
+  }
+
+  const std::vector<std::shared_ptr<std::function<void()>>>&
+  GetVoidFunctionPostHooks() const {
+    return void_function_post_hooks_;
+  }
+
  private:
   static const std::string& UnknownOpType() {
     static std::string kUnknownOpType{"unknown"};
@@ -203,6 +216,7 @@ class OpBase {
   // In order to reduce the compatibility phase
   // performance overhead, temporarily cache KernelContext
   static pten::KernelContext pt_kernel_context_;
+  std::vector<std::shared_ptr<std::function<void()>>> void_function_post_hooks_;
 };
 
 class GradOpNode {

paddle/fluid/pybind/imperative.cc

@@ -1640,6 +1640,26 @@ void BindImperative(py::module *m_ptr) {
                      "gradient or without gradient."));
              return self.GradVarBase()->RemoveVariableWrapperHook(hook_id);
            })
+      .def("_register_void_function_post_hook",
+           [](imperative::VarBase &self, const py::handle &hook) {
+             PADDLE_ENFORCE_EQ(
+                 !self.OverridedStopGradient() && self.HasGradVar(), true,
+                 platform::errors::InvalidArgument(
+                     "Cannot register void function post hook on a Tensor that "
+                     "stop "
+                     "gradient or without gradient."));
+             auto py_func = PyObjectCast<std::function<void()>>(hook.ptr());
+             VLOG(1) << 111;
+             auto grad_node = self.MutableGradVarBase()->GradNode();
+             VLOG(1) << 222;
+             VLOG(1) << (grad_node == nullptr);
+             for (auto &cur_op : *grad_node) {
+               VLOG(1) << 333;
+               cur_op.AddVoidFunctionPostHook(
+                   std::make_shared<std::function<void()>>(py_func));
+               VLOG(1) << 444;
+             }
+           })
       .def("_register_backward_hook",
            [](imperative::VarBase &self, const py::handle &hook) {
              PADDLE_ENFORCE_EQ(

python/paddle/fluid/tests/unittests/test_function_hook.py

+#   Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# 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.
+
+from __future__ import print_function
+
+import unittest
+import paddle
+import numpy as np
+
+import paddle.fluid.core as core
+from paddle import _C_ops
+
+
+class TestCapture:
+    def __init__(self):
+        self.list = []
+
+
+test_cap = TestCapture()
+
+
+def test_hook():
+    test_cap.list.append(1)
+
+
+def grad_hook(grad):
+    test_cap.list.append(2)
+
+    return grad
+
+
+class TestBakcwardFunctionHookError(unittest.TestCase):
+    def test_hook(self):
+        input_data = np.ones([4, 4]).astype('float32')
+
+        x = paddle.to_tensor(input_data.astype(np.float32), stop_gradient=False)
+        z = paddle.to_tensor(input_data.astype(np.float32), stop_gradient=False)
+
+        y = _C_ops.sigmoid(x)
+        out = _C_ops.matmul_v2(y, z, 'trans_x', False, 'trans_y', False)
+
+        out._register_void_function_post_hook(test_hook)
+        y._register_void_function_post_hook(test_hook)
+        y.register_hook(grad_hook)
+
+        out.backward()
+
+        assert test_cap.list == [1, 2, 1]
+
+
+if __name__ == "__main__":
+    unittest.main()