Cherry-pick PR 37420, fix inplace bug when the first grad_var(loss_gr…

…ad) is inplace var (#37420) * fix inplace bug * fix custom grad input error * add unittest * fix inplace bug
PaddlePaddle · Nov 23, 2021 · 5fd5c1c · 5fd5c1c · paddle-bot-old · Nov 24, 2021
1 parent f873d3a
commit 5fd5c1c
Show file tree

Hide file tree

Showing 2 changed files with 34 additions and 6 deletions.
diff --git a/paddle/fluid/imperative/basic_engine.cc b/paddle/fluid/imperative/basic_engine.cc
@@ -53,6 +53,10 @@ void BasicEngine::Init(
                     platform::errors::AlreadyExists(
                         "Accumulators are not empty before preparing it for "
                         "backward network execution."));
+  PADDLE_ENFORCE_EQ(accumulators_with_grad_node_.empty(), true,
+                    platform::errors::AlreadyExists(
+                        "Accumulators with grad_node as the key are not empty "
+                        "before preparing it for backward network execution."));
 
   for (size_t i = 0; i < tensors.size(); ++i) {
     auto var = tensors[i];
@@ -73,7 +77,6 @@ void BasicEngine::Init(
       VLOG(5) << "Clear the auto-grad graph from grad var " << var->Name()
               << " because of retain_graph=False when calling backward";
       var->GradVarBase()->SetGraphIsFreed(true);
-      var->GradVarBase()->ClearGradNode();
     }
 
     if (init_node == nullptr || var->OverridedStopGradient()) {
@@ -108,14 +111,18 @@ void BasicEngine::Init(
     }
 
     VariableWrapper* init_grad_var = var->GradVarBase()->SharedVar().get();
-    auto& accumulator = accumulators_[init_grad_var];
+    auto& accumulator =
+        accumulators_with_grad_node_[init_grad_var->GetGradNode()]
+                                    [init_grad_var];
     if (!accumulator) {
       if (FLAGS_sort_sum_gradient) {
         accumulator.reset(new SortedGradientAccumulator(init_grad_var));
       } else {
         accumulator.reset(new EagerGradientAccumulator(init_grad_var));
       }
     }
+    accumulator->IncreaseRefCnt();
+    accumulator->IncreaseCurCnt();
 
     init_nodes_.push_back(init_node);
   }
@@ -253,10 +260,6 @@ void BasicEngine::PrepareDeps() {
       node_deps_.empty(), true,
       platform::errors::AlreadyExists("Op deps are not empty before preparing "
                                       "it for backward network execution."));
-  PADDLE_ENFORCE_EQ(accumulators_with_grad_node_.empty(), true,
-                    platform::errors::AlreadyExists(
-                        "Accumulators with grad_node as the key are not empty "
-                        "before preparing it for backward network execution."));
 
   std::queue<GradOpNode*> q;
   std::unordered_set<GradOpNode*> visited;

diff --git a/python/paddle/fluid/tests/unittests/test_inplace.py b/python/paddle/fluid/tests/unittests/test_inplace.py
@@ -409,5 +409,30 @@ def inplace_api_processing(self, var):
         return var.subtract_(self.input_var_2)
 
 
+class TestLossIsInplaceVar(unittest.TestCase):
+    def test_loss_is_inplace_var(self):
+        with paddle.fluid.dygraph.guard():
+            var_a = paddle.ones((2, 2))
+            var_a.stop_gradient = False
+
+            var_b = var_a * 2
+            loss = var_b.tanh_()
+
+            loss.backward()
+            inplace_grad_var_a = var_a.grad.numpy()
+
+        with paddle.fluid.dygraph.guard():
+            var_a = paddle.ones((2, 2))
+            var_a.stop_gradient = False
+
+            var_b = var_a * 2
+            loss = var_b.tanh()
+
+            loss.backward()
+            grad_var_a = var_a.grad.numpy()
+
+        self.assertTrue(np.array_equal(inplace_grad_var_a, grad_var_a))
+
+
 if __name__ == '__main__':
     unittest.main()