updating repo

CHANGELOG.md

@@ -85,6 +85,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ### Changed
 
+- Fixed broken links in `HGBDataset` ([#7907](https://github.com/pyg-team/pytorch_geometric/pull/7907))
 - Fixed an issue where `SetTransformerAggregation` produced `NaN` values for isolates nodes ([#7902](https://github.com/pyg-team/pytorch_geometric/pull/7902))
 - Fixed `model_summary` on modules with uninitialized parameters ([#7884](https://github.com/pyg-team/pytorch_geometric/pull/7884))
 - Updated `QM9` data pre-processing to include the SMILES string ([#7867](https://github.com/pyg-team/pytorch_geometric/pull/7867))

test/transforms/test_rooted_subgraph.py

@@ -62,7 +62,7 @@ def test_rooted_rw_subgraph():
     assert out.n_sub_batch.tolist() == [0, 0, 1, 1, 2, 2]
 
 
-@withPackage('torch>=1.12.0')
+@withPackage('torch>=1.12.0', 'torch<2.1.0')
 def test_rooted_subgraph_minibatch():
     x = torch.randn(3, 8)
     edge_index = torch.tensor([[0, 1, 1, 2], [1, 0, 2, 1]])

test/utils/test_convert.py

@@ -462,13 +462,13 @@ def test_to_cugraph(edge_weight, directed, relabel_nodes):
     edge_list = graph.view_edge_list()
     assert edge_list is not None
 
-    edge_list = edge_list.sort_values(by=['src', 'dst'])
+    edge_list = edge_list.sort_values(by=[0, 1])
 
-    cu_edge_index = edge_list[['src', 'dst']].to_pandas().values
+    cu_edge_index = edge_list[[0, 1]].to_pandas().values
     cu_edge_index = torch.from_numpy(cu_edge_index).t()
     cu_edge_weight = None
     if edge_weight is not None:
-        cu_edge_weight = edge_list['weights'].to_pandas().values
+        cu_edge_weight = edge_list['2'].to_pandas().values
         cu_edge_weight = torch.from_numpy(cu_edge_weight)
 
     cu_edge_index, cu_edge_weight = sort_edge_index(cu_edge_index,

torch_geometric/datasets/hgb_dataset.py

@@ -40,20 +40,31 @@ class HGBDataset(InMemoryDataset):
             transformed version. The data object will be transformed before
             being saved to disk. (default: :obj:`None`)
     """
-
-    url = ('https://cloud.tsinghua.edu.cn/d/2d965d2fc2ee41d09def/files/'
-           '?p=%2F{}.zip&dl=1')
-
     names = {
         'acm': 'ACM',
         'dblp': 'DBLP',
         'freebase': 'Freebase',
         'imdb': 'IMDB',
     }
 
-    def __init__(self, root: str, name: str,
-                 transform: Optional[Callable] = None,
-                 pre_transform: Optional[Callable] = None):
+    urls = {
+        'acm': ('https://drive.google.com/uc?'
+                'export=download&id=1xbJ4QE9pcDJOcALv7dYhHDCPITX2Iddz'),
+        'dblp': ('https://drive.google.com/uc?'
+                 'export=download&id=1fLLoy559V7jJaQ_9mQEsC06VKd6Qd3SC'),
+        'freebase': ('https://drive.google.com/uc?'
+                     'export=download&id=1vw-uqbroJZfFsWpriC1CWbtHCJMGdWJ7'),
+        'imdb': ('https://drive.google.com/uc?'
+                 'export=download&id=18qXmmwKJBrEJxVQaYwKTL3Ny3fPqJeJ2'),
+    }
+
+    def __init__(
+        self,
+        root: str,
+        name: str,
+        transform: Optional[Callable] = None,
+        pre_transform: Optional[Callable] = None,
+    ):
         self.name = name.lower()
         assert self.name in set(self.names.keys())
         super().__init__(root, transform, pre_transform)
@@ -77,7 +88,7 @@ def processed_file_names(self) -> str:
         return 'data.pt'
 
     def download(self):
-        url = self.url.format(self.names[self.name])
+        url = self.urls[self.name]
         path = download_url(url, self.raw_dir)
         extract_zip(path, self.raw_dir)
         os.unlink(path)

torch_geometric/distributed/partition.py

@@ -150,10 +150,14 @@ def generate_partition(self):
                     size = (self.data[src].num_nodes, self.data[dst].num_nodes)
 
                     mask = part_data.edge_type == i
+                    rows = part_data.edge_index[0, mask]
+                    cols = part_data.edge_index[1, mask]
+                    global_rows = node_id[rows]
+                    global_cols = node_perm[cols]
                     out[edge_type] = {
                         'edge_id': edge_id[mask],
-                        'row': part_data.edge_index[0, mask],
-                        'col': part_data.edge_index[1, mask],
+                        'row': global_rows,
+                        'col': global_cols,
                         'size': size,
                     }
                 torch.save(out, osp.join(path, 'graph.pt'))
@@ -213,12 +217,16 @@ def generate_partition(self):
 
                 node_id = node_perm[start:end]
                 node_map[node_id] = pid
+                rows = part_data.edge_index[0]
+                cols = part_data.edge_index[1]
+                global_rows = node_id[rows]
+                global_cols = node_perm[cols]
 
                 torch.save(
                     {
                         'edge_id': edge_id,
-                        'row': part_data.edge_index[0],
-                        'col': part_data.edge_index[1],
+                        'row': global_rows,
+                        'col': global_cols,
                         'size': (data.num_nodes, data.num_nodes),
                     }, osp.join(path, 'graph.pt'))
 

torch_geometric/utils/convert.py

@@ -457,13 +457,13 @@ def from_cugraph(g: Any) -> Tuple[Tensor, Optional[Tensor]]:
     """
     df = g.view_edge_list()
 
-    src = from_dlpack(df['src'].to_dlpack()).long()
-    dst = from_dlpack(df['dst'].to_dlpack()).long()
+    src = from_dlpack(df[0].to_dlpack()).long()
+    dst = from_dlpack(df[1].to_dlpack()).long()
     edge_index = torch.stack([src, dst], dim=0)
 
     edge_weight = None
-    if 'weights' in df:
-        edge_weight = from_dlpack(df['weights'].to_dlpack())
+    if '2' in df:
+        edge_weight = from_dlpack(df['2'].to_dlpack())
 
     return edge_index, edge_weight