[Dataset] Contribute IGB-Homo dataset to node_classification.py

examples/graphbolt/node_classification.py

@@ -37,6 +37,7 @@
 │
 └───> All nodes set inference & Test set evaluation
 """
+
 import argparse
 import time
 
@@ -363,9 +364,17 @@ def parse_args():
         "--dataset",
         type=str,
         default="ogbn-products",
-        choices=["ogbn-arxiv", "ogbn-products", "ogbn-papers100M"],
+        choices=[
+            "ogbn-arxiv",
+            "ogbn-products",
+            "ogbn-papers100M",
+            "igb-hom-tiny",
+            "igb-hom-small",
+            "igb-hom-medium",
+        ],
         help="The dataset we can use for node classification example. Currently"
-        " ogbn-products, ogbn-arxiv, ogbn-papers100M datasets are supported.",
+        " ogbn-products, ogbn-arxiv, ogbn-papers100M and"
+        " igb-hom-[tiny|small|medium] datasets are supported.",
     )
     parser.add_argument(
         "--mode",

python/dgl/graphbolt/impl/ondisk_dataset.py

@@ -979,6 +979,16 @@
         .. note::
             Reverse edges are added to the original graph.
 
+    **igb-hom-[tiny|small|medium]**
+        The igb-hom-[tiny|small|medium] dataset is a homogeneous citation network,
+        which is designed for developers to train and evaluate GNN models with 
+        high fidelity. See more details in `igb-hom-[tiny|small|medium] 
+        <https://mlcommons.org/2024/06/gnn-for-mlperf-training-v4/>`_.
+
+        .. note::
+            Self edges are added to the original graph.
+            Node features are stored as float32.
+
     Parameters
     ----------
     name : str
@@ -1004,12 +1014,18 @@
         "ogbn-products-seeds",
         "ogbn-arxiv",
         "ogbn-arxiv-seeds",
+        "igb-hom-tiny",
+        "igb-hom-tiny-seeds",
+        "igb-hom-small",
+        "igb-hom-small-seeds",
     ]
     _large_datasets = [
         "ogb-lsc-mag240m",
         "ogb-lsc-mag240m-seeds",
         "ogbn-papers100M",
         "ogbn-papers100M-seeds",
+        "igb-hom-medium",
+        "igb-hom-medium-seeds",
     ]
     _all_datasets = _datasets + _large_datasets