fix bugs (dmlc#3008)

Co-authored-by: zhjwy9343 <[email protected]>

examples/pytorch/mvgrl/README.md

@@ -68,6 +68,7 @@ This example was implemented by [Hengrui Zhang](https://github.com/hengruizhang9
 --wd2              float   Weight decay of linear classifier.     Default is 0.0.
 --epsilon          float   Edge mask threshold.                   Default is 0.01.
 --hid_dim          int     Embedding dimension.                   Default is 512.
+--sample_size      int     Subgraph size.                         Default is 2000.
 ```
 
 ## How to run examples

examples/pytorch/mvgrl/node/main.py

@@ -126,4 +126,4 @@
         accs.append(acc * 100)
 
     accs = th.stack(accs)
-    print(accs.mean().item(), accs.std().item())
+    print(accs.mean().item(), accs.std().item())

examples/pytorch/mvgrl/node/main_sample.py

@@ -23,6 +23,7 @@
 parser.add_argument('--wd2', type=float, default=0., help='Weight decay of linear evaluator.')
 parser.add_argument('--epsilon', type=float, default=0.01, help='Edge mask threshold of diffusion graph.')
 parser.add_argument("--hid_dim", type=int, default=512, help='Hidden layer dim.')
+parser.add_argument("--sample_size", type=int, default=2000, help='Subgraph size.')
 
 args = parser.parse_args()
 
@@ -54,7 +55,7 @@
 
     n_node = graph.number_of_nodes()
 
-    sample_size = 2000
+    sample_size = args.sample_size
 
     lbl1 = th.ones(sample_size * 2)
     lbl2 = th.zeros(sample_size * 2)
@@ -153,4 +154,4 @@
         accs.append(acc * 100)
 
     accs = th.stack(accs)
-    print(accs.mean().item(), accs.std().item())
+    print(accs.mean().item(), accs.std().item())