[Example] Update GCN on ogbn-arxiv dataset (#2153)

* [Example] GCN on ogbn-arxiv dataset

* Add README.md

* Update GCN implementation on ogbn-arxiv

* Update GCN on ogbn-arxiv

* Fix typo

* Use evaluator to get results
Co-authored-by: Mufei Li <mufeili1996@gmail.com>
Co-authored-by: Zihao Ye <expye@outlook.com>

examples/pytorch/ogb/ogbn-arxiv/README.md

@@ -2,7 +2,7 @@
 
 Requires DGL 0.5 or later versions.
 
-Run `gcn.py` with `--use-linear` and `use-labels` enabled and you should directly see the result.
+Run `gcn.py` with `--use-linear` and `--use-labels` enabled and you should directly see the result.
 
 ```bash
 python3 gcn.py --use-linear --use-labels
@@ -17,12 +17,12 @@ usage: GCN on OGBN-Arxiv [-h] [--cpu] [--gpu GPU] [--n-runs N_RUNS] [--n-epochs
 
 optional arguments:
   -h, --help            show this help message and exit
-  --cpu                 CPU mode. This option overrides --gpu.
-  --gpu GPU             GPU device ID.
+  --cpu                 CPU mode. This option overrides --gpu. (default: False)
+  --gpu GPU             GPU device ID. (default: 0)
   --n-runs N_RUNS
   --n-epochs N_EPOCHS
-  --use-labels          Use labels in the training set as input features.
-  --use-linear          Use linear layers.
+  --use-labels          Use labels in the training set as input features. (default: False)
+  --use-linear          Use linear layer. (default: False)
   --lr LR
   --n-layers N_LAYERS
   --n-hidden N_HIDDEN
@@ -41,3 +41,4 @@ Here are the results over 10 runs.
 | Val acc    | 0.7361 ± 0.0009 | 0.7397 ± 0.0010 | 0.7399 ± 0.0008 |  0.7442 ± 0.0012  |
 | Test acc   | 0.7246 ± 0.0021 | 0.7270 ± 0.0016 | 0.7259 ± 0.0006 |  0.7306 ± 0.0024  |
 | Parameters |     109608      |     218152      |     119848      |      238632       |
+

examples/pytorch/ogb/ogbn-arxiv/gcn.py

@@ -11,7 +11,7 @@ import torch.nn.functional as F
 import torch.optim as optim
 from matplotlib import pyplot as plt
 from matplotlib.ticker import AutoMinorLocator, MultipleLocator
-from ogb.nodeproppred import DglNodePropPredDataset
+from ogb.nodeproppred import DglNodePropPredDataset, Evaluator
 
 from models import GCN
 
@@ -19,25 +19,38 @@ device = None
 in_feats, n_classes = None, None
 
 
-def compute_acc(pred, labels):
-    """
-    Compute the accuracy of prediction given the labels.
-    """
-    return (th.argmax(pred, dim=1) == labels).float().sum() / len(pred)
+def gen_model(args):
+    if args.use_labels:
+        model = GCN(
+            in_feats + n_classes, args.n_hidden, n_classes, args.n_layers, F.relu, args.dropout, args.use_linear
+        )
+    else:
+        model = GCN(in_feats, args.n_hidden, n_classes, args.n_layers, F.relu, args.dropout, args.use_linear)
+    return model
 
 
 def cross_entropy(x, labels):
-    y = F.cross_entropy(x, labels, reduction="none")
+    y = F.cross_entropy(x, labels[:, 0], reduction="none")
     y = th.log(0.5 + y) - math.log(0.5)
     return th.mean(y)
 
 
+def compute_acc(pred, labels, evaluator):
+    return evaluator.eval({"y_pred": pred.argmax(dim=-1, keepdim=True), "y_true": labels})["acc"]
+
+
 def add_labels(feat, labels, idx):
     onehot = th.zeros([feat.shape[0], n_classes]).to(device)
-    onehot[idx, labels[idx]] = 1
+    onehot[idx, labels[idx, 0]] = 1
     return th.cat([feat, onehot], dim=-1)
 
 
+def adjust_learning_rate(optimizer, lr, epoch):
+    if epoch <= 50:
+        for param_group in optimizer.param_groups:
+            param_group["lr"] = lr * epoch / 50
+
+
 def train(model, graph, labels, train_idx, optimizer, use_labels):
     model.train()
 
@@ -54,6 +67,7 @@ def train(model, graph, labels, train_idx, optimizer, use_labels):
     else:
         mask_rate = 0.5
         mask = th.rand(train_idx.shape) < mask_rate
+
         train_pred_idx = train_idx[mask]
 
     optimizer.zero_grad()
@@ -66,7 +80,7 @@ def train(model, graph, labels, train_idx, optimizer, use_labels):
 
 
 @th.no_grad()
-def evaluate(model, graph, labels, train_idx, val_idx, test_idx, use_labels):
+def evaluate(model, graph, labels, train_idx, val_idx, test_idx, use_labels, evaluator):
     model.eval()
 
     feat = graph.ndata["feat"]
@@ -80,32 +94,16 @@ def evaluate(model, graph, labels, train_idx, val_idx, test_idx, use_labels):
     test_loss = cross_entropy(pred[test_idx], labels[test_idx])
 
     return (
-        compute_acc(pred[train_idx], labels[train_idx]),
-        compute_acc(pred[val_idx], labels[val_idx]),
-        compute_acc(pred[test_idx], labels[test_idx]),
+        compute_acc(pred[train_idx], labels[train_idx], evaluator),
+        compute_acc(pred[val_idx], labels[val_idx], evaluator),
+        compute_acc(pred[test_idx], labels[test_idx], evaluator),
         train_loss,
         val_loss,
         test_loss,
     )
 
 
-def adjust_learning_rate(optimizer, lr, epoch):
-    if epoch <= 50:
-        for param_group in optimizer.param_groups:
-            param_group["lr"] = lr * epoch / 50
-
-
-def gen_model(args):
-    if args.use_labels:
-        model = GCN(
-            in_feats + n_classes, args.n_hidden, n_classes, args.n_layers, F.relu, args.dropout, args.use_linear
-        )
-    else:
-        model = GCN(in_feats, args.n_hidden, n_classes, args.n_layers, F.relu, args.dropout, args.use_linear)
-    return model
-
-
-def run(args, graph, labels, train_idx, val_idx, test_idx, n_running):
+def run(args, graph, labels, train_idx, val_idx, test_idx, evaluator, n_running):
     # define model and optimizer
     model = gen_model(args)
     model = model.to(device)
@@ -128,27 +126,28 @@ def run(args, graph, labels, train_idx, val_idx, test_idx, n_running):
         adjust_learning_rate(optimizer, args.lr, epoch)
 
         loss, pred = train(model, graph, labels, train_idx, optimizer, args.use_labels)
-        acc = compute_acc(pred[train_idx], labels[train_idx])
+        acc = compute_acc(pred[train_idx], labels[train_idx], evaluator)
 
         train_acc, val_acc, test_acc, train_loss, val_loss, test_loss = evaluate(
-            model, graph, labels, train_idx, val_idx, test_idx, args.use_labels
+            model, graph, labels, train_idx, val_idx, test_idx, args.use_labels, evaluator
         )
 
         lr_scheduler.step(loss)
 
-        toc = time.time()
-        total_time += toc - tic
+        train_acc, val_acc, test_acc, train_loss, val_loss, test_loss = evaluate(
+            model, graph, labels, train_idx, val_idx, test_idx, args.use_labels
+        )
 
         # if val_acc > best_val_acc:
         if val_loss < best_val_loss:
-            best_val_loss = val_loss.item()
-            best_val_acc = val_acc.item()
-            best_test_acc = test_acc.item()
+            best_val_loss = val_loss
+            best_val_acc = val_acc
+            best_test_acc = test_acc
 
         if epoch % args.log_every == 0:
             print(f"Epoch: {epoch}/{args.n_epochs}")
             print(
-                f"Loss: {loss.item():.4f}, Acc: {acc.item():.4f}\n"
+                f"Loss: {loss.item():.4f}, Acc: {acc:.4f}\n"
                 f"Train/Val/Test loss: {train_loss:.4f}/{val_loss:.4f}/{test_loss:.4f}\n"
                 f"Train/Val/Test/Best val/Best test acc: {train_acc:.4f}/{val_acc:.4f}/{test_acc:.4f}/{best_val_acc:.4f}/{best_test_acc:.4f}"
             )
@@ -234,10 +233,11 @@ def main():
 
     # load data
     data = DglNodePropPredDataset(name="ogbn-arxiv")
+    evaluator = Evaluator(name="ogbn-arxiv")
+
     splitted_idx = data.get_idx_split()
     train_idx, val_idx, test_idx = splitted_idx["train"], splitted_idx["valid"], splitted_idx["test"]
     graph, labels = data[0]
-    labels = labels[:, 0]
 
     # add reverse edges
     srcs, dsts = graph.all_edges()
@@ -263,7 +263,7 @@ def main():
     test_accs = []
 
     for i in range(args.n_runs):
-        val_acc, test_acc = run(args, graph, labels, train_idx, val_idx, test_idx, i)
+        val_acc, test_acc = run(args, graph, labels, train_idx, val_idx, test_idx, evaluator, i)
         val_accs.append(val_acc)
         test_accs.append(test_acc)