Update

examples/ngcf.py

@@ -96,6 +96,7 @@ def sparse_dropout(self, row: Tensor, col: Tensor, value: Tensor,
                        rate: float, nnz: int) -> SparseTensor:
         rand = 1 - rate
         rand += torch.rand(nnz)
+        assert isinstance(rand, Tensor)
         dropout_mask = torch.floor(rand).type(torch.bool)
         adj = SparseTensor(
             row=row[dropout_mask],
@@ -105,9 +106,10 @@ def sparse_dropout(self, row: Tensor, col: Tensor, value: Tensor,
         )
         return adj
 
-    def get_embedding(self, norm_adj: Tensor, device=torch.device) -> Tensor:
+    def get_embedding(self, norm_adj: SparseTensor,
+                      device=torch.device) -> Tensor:
         ego_emb = self.emb.weight
-        all_embs = [ego_emb]
+        all_embs: List[Tensor] = [ego_emb]
         if self.node_dropout > 0 and self.training:
             row, col, value = norm_adj.coo()
             adj = self.sparse_dropout(
@@ -129,7 +131,7 @@ def get_embedding(self, norm_adj: Tensor, device=torch.device) -> Tensor:
             ego_emb = F.dropout(ego_emb)
             norm_emb = F.normalize(ego_emb, p=2, dim=1)
             all_embs += [norm_emb]
-        all_embs = torch.cat(all_embs, 1)
+        all_embs: Tensor = torch.cat(all_embs, 1)
         return all_embs
 
     def recommendation_loss(