solve the problem of repeated definition of dropout in forward of MLP

Author: gyzhou2000

gammagl/models/gin.py

@@ -49,14 +49,15 @@ def __init__(self, in_channels,
 
         self.mlp = MLP([hidden_channels, hidden_channels, out_channels],
                        norm=None, dropout=0.5)
+        self.relu = tlx.ReLU()
 
     def forward(self, x, edge_index, batch):
         if x is None:
             # x = tlx.ones((batch.shape[0], 1), dtype=tlx.float32)
             x = tlx.random_normal((batch.shape[0], 1), dtype=tlx.float32)
 
         for conv in self.convs:
-            x = tlx.relu(conv(x, edge_index))
+            x = self.relu(conv(x, edge_index))
 
         x = global_sum_pool(x, batch)
         return self.mlp(x)

gammagl/models/mlp.py

@@ -46,6 +46,9 @@ def __init__(self,
             dropout[-1] = 0.
         assert len(dropout) == len(channel_list) - 1
         self.dropout = dropout
+        self.dropouts = tlx.nn.ModuleList()
+        for i in range(len(dropout)):
+            self.dropouts.append(tlx.nn.Dropout(p=dropout[i]))
 
         if isinstance(bias, bool):
             bias = [bias] * (len(channel_list) - 1)
@@ -89,12 +92,14 @@ def forward(self, x, return_emb=None):
             if self.act is not None and not self.act_first:
                 x = self.act(x)
 
-            x = tlx.nn.Dropout(p=self.dropout[i])(x)
+            # x = tlx.nn.Dropout(p=self.dropout[i])(x)
+            x = self.dropouts[i](x)
             emb = x
 
         if self.plain_last:
             x = self.lins[-1](x)
-            x = tlx.nn.Dropout(p=self.dropout[-1])(x)
+            # x = tlx.nn.Dropout(p=self.dropout[-1])(x)
+            x = self.dropouts[-1](x)
 
         return (x, emb) if isinstance(return_emb, bool) else x
 

gammagl/mpops/torch_ext/cpu/segment_max_cpu.cpp

@@ -34,7 +34,7 @@ std::tuple<torch::Tensor, torch::Tensor> segment_max_cpu_forward(
   auto index_data = index.data_ptr<int64_t>();
   auto arg_out_data = arg_out.data_ptr<int64_t>();
 
-  AT_DISPATCH_ALL_TYPES_AND2(at::ScalarType::Half, at::ScalarType::BFloat16, x.scalar_type(), "segment_mean_cpu_forward", [&]() {
+  AT_DISPATCH_ALL_TYPES_AND2(at::ScalarType::Half, at::ScalarType::BFloat16, x.scalar_type(), "segment_max_cpu_forward", [&]() {
       out.fill_(std::numeric_limits<scalar_t>::lowest());
       auto x_data = x.data_ptr<scalar_t>();
       auto out_data = out.data_ptr<scalar_t>();