Improving the HARD_GAT example (#6036)

Co-authored-by: Hongzhi (Steve), Chen <chenhongzhi.nkcs@gmail.com>

examples/pytorch/hardgat/train.py

@@ -64,7 +64,7 @@ def main(args):
     val_mask = g.ndata["val_mask"]
     test_mask = g.ndata["test_mask"]
     num_feats = features.shape[1]
-    n_classes = data.num_labels
+    n_classes = data.num_classes
     n_edges = g.num_edges()
     print(
         """----Data statistics------'
@@ -115,7 +115,7 @@ def main(args):
     )
 
     # initialize graph
-    dur = []
+    mean = 0
     for epoch in range(args.epochs):
         model.train()
         if epoch >= 3:
@@ -129,8 +129,7 @@ def main(args):
         optimizer.step()
 
         if epoch >= 3:
-            dur.append(time.time() - t0)
-
+            mean = (mean * (epoch - 3) + (time.time() - t0)) / (epoch - 2)
             train_acc = accuracy(logits[train_mask], labels[train_mask])
 
             if args.fastmode:
@@ -145,11 +144,11 @@ def main(args):
                 "Epoch {:05d} | Time(s) {:.4f} | Loss {:.4f} | TrainAcc {:.4f} |"
                 " ValAcc {:.4f} | ETputs(KTEPS) {:.2f}".format(
                     epoch,
-                np.mean(dur),
+                    mean,
                     loss.item(),
                     train_acc,
                     val_acc,
-                n_edges / np.mean(dur) / 1000,
+                    n_edges / mean / 1000,
                 )
             )