Revert "[CI] Temmprarily skip ogb-related tests" (#5944)

tests/python/common/data/test_data.py

@@ -1789,7 +1789,6 @@ def test_as_nodepred2():
     assert len(ds.train_idx) == int(ds[0].num_nodes("Personen") * 0.1)
 
 
-@unittest.skip("ogb is not available")
 @unittest.skipIf(
     dgl.backend.backend_name != "pytorch", reason="ogb only supports pytorch"
 )
@@ -1842,7 +1841,6 @@ def test_as_linkpred():
     assert 4000 < ds.test_edges[1][0].shape[0] <= 4224
 
 
-@unittest.skip("ogb is not available")
 @unittest.skipIf(
     dgl.backend.backend_name != "pytorch", reason="ogb only supports pytorch"
 )
@@ -2076,7 +2074,6 @@ def test_as_graphpred_reprocess():
     assert len(ds.train_idx) == int(len(ds) * 0.1)
 
 
-@unittest.skip("ogb is not available")
 @unittest.skipIf(
     dgl.backend.backend_name != "pytorch", reason="ogb only supports pytorch"
 )

tutorials/large/L1_large_node_classification.py

@@ -24,7 +24,7 @@ Sampling for GNN Training <L0_neighbor_sampling_overview>`.
 #
 # OGB already prepared the data as DGL graph.
 #
-exit(0)
+
 import os
 
 os.environ["DGLBACKEND"] = "pytorch"

tutorials/large/L2_large_link_prediction.py

@@ -51,7 +51,7 @@ Sampling for Node Classification <L1_large_node_classification>`.
 # This tutorial loads the dataset from the ``ogb`` package as in the
 # :doc:`previous tutorial <L1_large_node_classification>`.
 #
-exit(0)
+
 import os
 
 os.environ["DGLBACKEND"] = "pytorch"

tutorials/large/L4_message_passing.py

@@ -12,7 +12,7 @@ for stochastic GNN training. It assumes that
    works <L1_large_node_classification>`.
 
 """
-exit(0)
+
 import os
 
 os.environ["DGLBACKEND"] = "pytorch"

tutorials/multi/2_node_classification.py

@@ -32,7 +32,7 @@ models with multi-GPU with ``DistributedDataParallel``.
 # Classification <../large/L1_large_node_classification>`
 # tutorial.
 #
-exit(0)
+
 import os
 
 os.environ["DGLBACKEND"] = "pytorch"