[NN] Fix GINConv (#3692)

* Update

* Update

* Fix

* Update

* Update

* Update

* Update

* Fix

* Update

* Update

* Update

* Update

* Fix lint

* lint

* Update

* Update

* lint fix

* Fix CI

* Fix

* Fix CI

* Update

* Fix

* Update

* Update

* Update ginconv.py

* Update test_nn.py
Co-authored-by: Ubuntu <ubuntu@ip-172-31-31-136.us-west-2.compute.internal>

python/dgl/nn/pytorch/conv/ginconv.py

@@ -34,9 +34,9 @@ class GINConv(nn.Module):
     ----------
     apply_func : callable activation function/layer or None
         If not None, apply this function to the updated node feature,
-        the :math:`f_\Theta` in the formula.
+        the :math:`f_\Theta` in the formula, default: None.
     aggregator_type : str
-        Aggregator type to use (``sum``, ``max`` or ``mean``).
+        Aggregator type to use (``sum``, ``max`` or ``mean``), default: 'sum'.
     init_eps : float, optional
         Initial :math:`\epsilon` value, default: ``0``.
     learn_eps : bool, optional
@@ -90,8 +90,8 @@ class GINConv(nn.Module):
              0.0000]], grad_fn=<ReluBackward0>)
     """
     def __init__(self,
-                 apply_func,
-                 aggregator_type,
+                 apply_func=None,
+                 aggregator_type='sum',
                  init_eps=0,
                  learn_eps=False,
                  activation=None):
@@ -108,22 +108,6 @@ class GINConv(nn.Module):
         else:
             self.register_buffer('eps', th.FloatTensor([init_eps]))
 
-        self.reset_parameters()
-
-    def reset_parameters(self):
-        r"""
-
-        Description
-        -----------
-        Reinitialize learnable parameters.
-
-        Note
-        ----
-        The model parameters are initialized using Glorot uniform initialization.
-        """
-        gain = nn.init.calculate_gain('relu')
-        nn.init.xavier_normal_(self.apply_func.weight, gain=gain)
-
     def forward(self, graph, feat, edge_weight=None):
         r"""
 

tests/pytorch/test_nn.py

@@ -789,6 +789,11 @@ def test_gin_conv(g, idtype, aggregator_type):
 
     assert h.shape == (g.number_of_dst_nodes(), 12)
     
+    gin = nn.GINConv(None, aggregator_type)
+    th.save(gin, tmp_buffer)
+    gin = gin.to(ctx)
+    h = gin(g, feat)
+
 @parametrize_dtype
 @pytest.mark.parametrize('g', get_cases(['bipartite'], exclude=['zero-degree']))
 @pytest.mark.parametrize('aggregator_type', ['mean', 'max', 'sum'])