fix(mge/imperative): fix tracer leak

GitOrigin-RevId: 1a8ac20b46e25897a34c62da088dbf49e0e19ee6

imperative/python/megengine/core/autodiff/grad.py

@@ -82,6 +82,7 @@ class Grad:
         # ops forms the computational graph
         self.ops = []
 
+        self._attached_tensors = weakref.WeakSet()
         self._enabled = True
 
     @property
@@ -107,6 +108,7 @@ class Grad:
         return self
 
     def _new_variable(self, owner, opnode=None, callback=None):
+        self._attached_tensors.add(owner)
         return VariableNode(self, owner, opnode=opnode, callback=callback)
 
     def _new_opnode(self, inputs, outputs):
@@ -131,13 +133,18 @@ class Grad:
     def __enter__(self):
         return self
 
-    def __exit__(self, *_):
+    def _exit(self):
         """clear all resources"""
         self._enabled = False
         for o in self.ops:
             o = o()
             if o:
                 o.clear()
+        for i in self._attached_tensors:
+            i._extra_data.pop(self, None)
+
+    def __exit__(self, *_):
+        self._exit()
 
     def __call__(self, ys, dys):
         """ Defines Grad().
@@ -275,6 +282,11 @@ class Grad:
         for v in cache.values():
             assert v is None
 
+        self._exit()
+
+    def __del__(self):
+        self._exit()
+
 
 class clearable:
     __cleared = False

imperative/python/test/unit/core/test_autodiff.py

@@ -6,6 +6,7 @@
 # Unless required by applicable law or agreed to in writing,
 # software distributed under the License is distributed on an
 # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+import gc
 import platform
 import weakref
 
@@ -158,6 +159,37 @@ def test_grad_with_tensor_wrapper():
     np.testing.assert_almost_equal(x.grad.numpy(), 4 * x_np ** 3, decimal=6)
 
 
+def test_release():
+    def check(f):
+        n = 0
+        d = None
+        for i in range(3):
+            f()
+            m = len(gc.get_objects())
+            d = m - n
+            n = m
+        assert d == 0
+
+    x = TensorWrapper([0.0])
+    dy = TensorWrapper(np.ones_like(x.numpy()))
+
+    @check
+    def _():
+        g = Grad().wrt(x)
+        y = x * x
+        g(y, dy)
+
+    @check
+    def _():
+        with Grad().wrt(x) as g:
+            pass
+
+    @check
+    def _():
+        with Grad().wrt(x) as g:
+            y = x * x
+
+
 def test_grad_inplace():
     x_np = np.random.rand(10).astype("float32")
     x = TensorWrapper(x_np)