add thumbnails for tutorials (#2695)

Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>
Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

python/dgl/nn/pytorch/conv/cfconv.py

@@ -134,7 +134,11 @@ class CFConv(nn.Module):
             Updated node representations.
         """
         with g.local_scope():
-            g.ndata['hv'] = self.project_node(node_feats)
+            if isinstance(node_feats, tuple):
+                node_feats_src, _ = node_feats
+            else:
+                node_feats_src = node_feats
+            g.srcdata['hv'] = self.project_node(node_feats_src)
             g.edata['he'] = self.project_edge(edge_feats)
             g.update_all(fn.u_mul_e('hv', 'he', 'm'), fn.sum('m', 'h'))
-            return self.project_out(g.ndata['h'])
+            return self.project_out(g.dstdata['h'])

tutorials/blitz/1_introduction.py

@@ -215,3 +215,6 @@ train(g, model)
 # -  :ref:`The list of datasets provided by DGL <apidata>`.
 # 
 
+
+# Thumbnail Courtesy: Stanford CS224W Notes
+# sphinx_gallery_thumbnail_path = '_static/blitz_1_introduction.png'

tutorials/blitz/2_dglgraph.py

@@ -225,3 +225,6 @@ print(sg2)
 #    :func:`dgl.load_graphs`
 # 
 
+
+# Thumbnail Courtesy: Wikipedia
+# sphinx_gallery_thumbnail_path = '_static/blitz_2_dglgraph.png'

tutorials/blitz/3_message_passing.py

@@ -353,3 +353,6 @@ def sum_udf(nodes):
 #    Code <guide-message-passing-efficient>`.
 # 
 
+
+# Thumbnail Courtesy: Representation Learning on Networks, Jure Leskovec, WWW 2018
+# sphinx_gallery_thumbnail_path = '_static/blitz_3_message_passing.png'

tutorials/blitz/4_link_predict.py

@@ -339,3 +339,6 @@ with torch.no_grad():
     neg_score = pred(test_neg_g, h)
     print('AUC', compute_auc(pos_score, neg_score))
 
+
+# Thumbnail Courtesy: Link Prediction with Neo4j, Mark Needham
+# sphinx_gallery_thumbnail_path = '_static/blitz_4_link_predict.png'

tutorials/blitz/5_graph_classification.py

@@ -207,3 +207,6 @@ print('Test accuracy:', num_correct / num_tests)
 #    for an end-to-end graph classification model.
 # 
 
+
+# Thumbnail Courtesy: DGL
+# sphinx_gallery_thumbnail_path = '_static/blitz_5_graph_classification.png'

tutorials/blitz/6_load_data.py

@@ -222,3 +222,6 @@ dataset = SyntheticDataset()
 graph, label = dataset[0]
 print(graph, label)
 
+
+# Thumbnail Courtesy: (Un)common Use Cases for Graph Databases, Michal Bachman
+# sphinx_gallery_thumbnail_path = '_static/blitz_6_load_data.png'

tutorials/large/L0_neighbor_sampling_overview.py

@@ -113,3 +113,6 @@ By the end of this tutorial, you will be able to
 # DGL <L1_large_node_classification>`
 #
 
+
+# Thumbnail Courtesy: Understanding graph embedding methods and their applications, Mengjia Xu
+# sphinx_gallery_thumbnail_path = '_static/large_L0_neighbor_sampling_overview.png'

tutorials/large/L1_large_node_classification.py

@@ -334,4 +334,5 @@ for epoch in range(10):
 #
 
 
-
+# Thumbnail Courtesy: Stanford CS224W Notes
+# sphinx_gallery_thumbnail_path = '_static/blitz_1_introduction.png'

tutorials/large/L2_large_link_prediction.py

@@ -364,3 +364,6 @@ for epoch in range(1):
 # for link prediction with neighbor sampling.
 #
 
+
+# Thumbnail Courtesy: Link Prediction with Neo4j, Mark Needham
+# sphinx_gallery_thumbnail_path = '_static/blitz_4_link_predict.png'

tutorials/large/L4_message_passing.py

@@ -387,3 +387,6 @@ class SAGEConvForBoth(nn.Module):
             h_total = torch.cat([h_dst, h_N], dim=1)
             return self.linear(h_total)
 
+
+# Thumbnail Courtesy: Representation Learning on Networks, Jure Leskovec, WWW 2018
+# sphinx_gallery_thumbnail_path = '_static/blitz_3_message_passing.png'