[Dataset] Update ogb arxiv & Fix epochs (#294)

THUDM · Oct 28, 2021 · e34f305 · e34f305
1 parent 0a64b59
commit e34f305
Show file tree

Hide file tree

Showing 53 changed files with 507 additions and 1,207 deletions.
diff --git a/README.md b/README.md
@@ -88,7 +88,7 @@ from cogdl import experiment
 experiment(dataset="cora", model="gcn")
 
 # set other hyper-parameters
-experiment(dataset="cora", model="gcn", hidden_size=32, max_epoch=200)
+experiment(dataset="cora", model="gcn", hidden_size=32, epochs=200)
 
 # run over multiple models on different seeds
 experiment(dataset="cora", model=["gcn", "gat"], seed=[1, 2])

diff --git a/README_CN.md b/README_CN.md
@@ -86,7 +86,7 @@ from cogdl import experiment
 experiment(dataset="cora", model="gcn")
 
 # set other hyper-parameters
-experiment(dataset="cora", model="gcn", hidden_size=32, max_epoch=200)
+experiment(dataset="cora", model="gcn", hidden_size=32, epochs=200)
 
 # run over multiple models on different seeds
 experiment(dataset="cora", model=["gcn", "gat"], seed=[1, 2])

diff --git a/cogdl/configs.py b/cogdl/configs.py
@@ -3,7 +3,7 @@
         "chebyshev": {"general": {}},
         "dropedge_gcn": {"general": {}},
         "gat": {
-            "general": {"lr": 0.005, "max_epoch": 1000},
+            "general": {"lr": 0.005, "epochs": 1000},
             "citeseer": {"weight_decay": 0.001},
             "pubmed": {"weight_decay": 0.001},
             "ppi-large": {
@@ -24,7 +24,7 @@
                 "hidden_size": 2048,
                 "num_layers": 2,
                 "dropout": 0.3,
-                "max_epoch": 7000,
+                "epochs": 7000,
                 # 78.04
             },
             "flickr": {
@@ -39,7 +39,7 @@
         },
         "gcnii": {
             "general": {
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "dropout": 0.5,
                 "wd1": 0.001,
                 "wd2": 5e-4,
@@ -54,7 +54,7 @@
                 "hidden_size": 256,
                 "lr": 0.001,
                 "patience": 200,
-                "max_epoch": 2000,
+                "epochs": 2000,
                 "lmbda": 0.6,
                 "dropout": 0.7,
             },
@@ -73,7 +73,7 @@
                 "wd2": 0,
                 "lr": 0.001,
                 "residual": True,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "lmbda": 1.0,
                 # 96.42
             },
@@ -93,7 +93,7 @@
         },
         "grand": {
             "general": {
-                "max_epoch": 1000,
+                "epochs": 1000,
             },
             "cora": {
                 "order": 8,
@@ -139,7 +139,7 @@
         "sgcpn": {
             "general": {
                 "lr": 0.005,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "patience": 1000,
                 "norm_mode": "PN",
                 "norm_scale": 10,
@@ -157,14 +157,14 @@
         "srgcn": {
             "general": {
                 "lr": 0.005,
-                "max_epoch": 1000,
+                "epochs": 1000,
             },
             "cora": {"dropout": 0.6},
             "citeseer": {"dropout": 0.6},
         },
         "unet": {
             "general": {
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "n_dropout": 0.90,
                 "adj_dropout": 0.05,
                 "hidden_size": 128,
@@ -182,7 +182,7 @@
                 "dropout": 0.3,
                 "num_layers": 2,
                 "hidden_size": 256,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "num_iterations": 4,
                 "alpha": 0.5,
                 # 0.5227
@@ -193,7 +193,7 @@
                 "alpha": 0.1087,
                 "num_iterations": 7,
                 "num_layers": 2,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "hidden_size": 32,
                 "dropout": 0.8185,
                 #
@@ -206,7 +206,7 @@
                 "alpha": 0.171,
                 "weight_decay": 0.0004117,
                 "num_layers": 2,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 # 0.7199
             },
             "pubmed": {
@@ -215,7 +215,7 @@
                 "dropout": 0.3766,
                 "alpha": 0.1201,
                 "weight_decay": 0.0002754,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "lr": 0.001,
             },
             "reddit": {
@@ -289,7 +289,7 @@
             "general": {
                 "hidden_size": 128,
                 "lr": 0.001,
-                "max_epoch": 500,
+                "epochs": 500,
                 "hidden_size1": 1000,
                 "hidden_size2": 128,
                 "noise": 0.2,
@@ -342,7 +342,7 @@
         "grace": {
             "general": {
                 "weight_decay": 0,
-                "max_epoch": 1000,
+                "epochs": 1000,
                 "patience": 20,
             },
             "cora": {
@@ -351,7 +351,7 @@
                 "tau": 0.4,
                 "drop_feature_rates": [0.3, 0.4],
                 "drop_edge_rates": [0.2, 0.4],
-                "max_epoch": 200,
+                "epochs": 200,
                 "hidden_size": 128,
                 "proj_hidden_size": 128,
             },
@@ -378,7 +378,7 @@
         "unsup_graphsage": {
             "lr": 0.001,
             "weight_decay": 0,
-            "max_epoch": 3000,
+            "epochs": 3000,
         },
         "revgcn": {
             "general": {},
@@ -417,7 +417,7 @@
             },
             "mutag": {
                 "num_layers": 1,
-                "epoch": 20,
+                "epochs": 20,
             },
             "imdb-b": {"degree_feature": True},
             "imdb-m": {"degree_feature": True},
@@ -449,7 +449,7 @@
             "nci1": {
                 "lr": 0.001,
                 "window_size": 8,
-                "epoch": 10,
+                "epochs": 10,
                 "iteration": 4,
             },
             "reddit-b": {

diff --git a/cogdl/data/dataset.py b/cogdl/data/dataset.py
@@ -170,7 +170,7 @@ def num_graphs(self):
         return 1
 
     def __repr__(self):  # pragma: no cover
-        return "{}({})".format(self.__class__.__name__, len(self))
+        return "{}".format(self.__class__.__name__)
 
 
 class MultiGraphDataset(Dataset):
@@ -268,3 +268,6 @@ def __getitem__(self, item):
 
     def __len__(self):
         return len(self.data)
+
+    def __repr__(self):  # pragma: no cover
+        return "{}({})".format(self.__class__.__name__, len(self))
diff --git a/cogdl/datasets/ogb.py b/cogdl/datasets/ogb.py
@@ -39,7 +39,7 @@ def process(self):
         name = self.name.replace("_", "-")
         dataset = NodePropPredDataset(name, self.root)
         graph, y = dataset[0]
-        x = torch.tensor(graph["node_feat"]) if graph["node_feat"] is not None else None
+        x = torch.tensor(graph["node_feat"]).contiguous() if graph["node_feat"] is not None else None
         y = torch.tensor(y.squeeze())
         row, col = graph["edge_index"][0], graph["edge_index"][1]
         row = torch.from_numpy(row)
@@ -75,17 +75,6 @@ def __init__(self, data_path="data"):
         dataset = "ogbn-arxiv"
         super(OGBArxivDataset, self).__init__(data_path, dataset)
 
-    def get_evaluator(self):
-        evaluator = NodeEvaluator(name="ogbn-arxiv")
-
-        def wrap(y_pred, y_true):
-            y_pred = y_pred.argmax(dim=-1, keepdim=True)
-            y_true = y_true.view(-1, 1)
-            input_dict = {"y_true": y_true, "y_pred": y_pred}
-            return evaluator.eval(input_dict)["acc"]
-
-        return wrap
-
 
 class OGBProductsDataset(OGBNDataset):
     def __init__(self, data_path="data"):

diff --git a/cogdl/experiments.py b/cogdl/experiments.py
@@ -3,6 +3,7 @@
 import os
 import inspect
 from collections import defaultdict, namedtuple
+import warnings
 
 import torch
 import torch.nn as nn
@@ -177,7 +178,7 @@ def train(args):  # noqa: C901
         lr=args.lr,
         weight_decay=args.weight_decay,
         n_warmup_steps=args.n_warmup_steps,
-        max_epoch=args.max_epoch,
+        epochs=args.epochs,
         batch_size=args.batch_size if hasattr(args, "batch_size") else 0,
     )
 
@@ -194,7 +195,7 @@ def train(args):  # noqa: C901
 
     # setup controller
     trainer = Trainer(
-        max_epoch=args.max_epoch,
+        epochs=args.epochs,
         device_ids=args.devices,
         cpu=args.cpu,
         save_emb_path=args.save_emb_path,
@@ -285,9 +286,16 @@ def experiment(dataset, model, **kwargs):
         for key, value in kwargs.items():
             if key != "args":
                 args.__setattr__(key, value)
+    if isinstance(model[0], nn.Module):
+        args.model = [x.model_name for x in model]
+    print(args)
     args.dataset = dataset
     args.model = model
 
+    if args.max_epoch is not None:
+        warnings.warn("The max_epoch is deprecated and will be removed in the future, please use epochs instead!")
+        args.epochs = args.max_epoch
+
     if "search_space" in kwargs:
         return auto_experiment(args)
 

diff --git a/cogdl/models/emb/dgk.py b/cogdl/models/emb/dgk.py
@@ -18,7 +18,7 @@ class DeepGraphKernel(BaseModel):
         window (int) : The actual context size which is considered in language model.
         sampling_rate (float) : Parameter in word2vec.
         iteration (int) : The number of iteration in WL method.
-        epoch (int) : The number of training iteration.
+        epochs (int) : The number of training iteration.
         alpha (float) : The learning rate of word2vec.
     """
 
@@ -29,13 +29,13 @@ def add_args(parser):
         parser.add_argument("--min-count", type=int, default=1)
         parser.add_argument("--sampling", type=float, default=0.0001)
         parser.add_argument("--iteration", type=int, default=2)
-        parser.add_argument("--epoch", type=int, default=20)
+        parser.add_argument("--epochs", type=int, default=20)
         parser.add_argument("--alpha", type=float, default=0.01)
 
     @classmethod
     def build_model_from_args(cls, args):
         return cls(
-            args.hidden_size, args.min_count, args.window_size, args.sampling, args.iteration, args.epoch, args.alpha
+            args.hidden_size, args.min_count, args.window_size, args.sampling, args.iteration, args.epochs, args.alpha
         )
 
     @staticmethod
@@ -67,7 +67,7 @@ def wl_iterations(graph, features, rounds):
             features = new_feats
         return wl_features
 
-    def __init__(self, hidden_dim, min_count, window_size, sampling_rate, rounds, epoch, alpha, n_workers=4):
+    def __init__(self, hidden_dim, min_count, window_size, sampling_rate, rounds, epochs, alpha, n_workers=4):
         super(DeepGraphKernel, self).__init__()
         self.hidden_dim = hidden_dim
         self.min_count = min_count
@@ -77,7 +77,7 @@ def __init__(self, hidden_dim, min_count, window_size, sampling_rate, rounds, ep
         self.rounds = rounds
         self.model = None
         self.gl_collections = None
-        self.epoch = epoch
+        self.epochs = epochs
         self.alpha = alpha
 
     def forward(self, graphs, **kwargs):
@@ -93,7 +93,7 @@ def forward(self, graphs, **kwargs):
             min_count=self.min_count,
             sample=self.sampling_rate,
             workers=self.n_workers,
-            iter=self.epoch,
+            iter=self.epochs,
             alpha=self.alpha,
         )
         vectors = np.asarray([model.wv[str(node)] for node in model.wv.index2word])

diff --git a/cogdl/models/emb/dngr.py b/cogdl/models/emb/dngr.py
@@ -48,7 +48,7 @@ class DNGR(BaseModel):
         noise (float) : Denoise rate of DAE.
         alpha (float) : Parameter in DNGR.
         step (int) : The max step in random surfing.
-        max_epoch (int) : The max epoches in training step.
+        epochs (int) : The max epoches in training step.
         lr (float) : Learning rate in DNGR.
     """
 
@@ -67,17 +67,17 @@ def add_args(parser):
     @classmethod
     def build_model_from_args(cls, args):
         return cls(
-            args.hidden_size1, args.hidden_size2, args.noise, args.alpha, args.step, args.max_epoch, args.lr, args.cpu
+            args.hidden_size1, args.hidden_size2, args.noise, args.alpha, args.step, args.epochs, args.lr, args.cpu
         )
 
-    def __init__(self, hidden_size1, hidden_size2, noise, alpha, step, max_epoch, lr, cpu):
+    def __init__(self, hidden_size1, hidden_size2, noise, alpha, step, epochs, lr, cpu):
         super(DNGR, self).__init__()
         self.hidden_size1 = hidden_size1
         self.hidden_size2 = hidden_size2
         self.noise = noise
         self.alpha = alpha
         self.step = step
-        self.max_epoch = max_epoch
+        self.epochs = epochs
         self.lr = lr
         self.cpu = cpu
 
@@ -144,7 +144,7 @@ def forward(self, graph, return_dict=False):
         opt = torch.optim.Adam(model.parameters(), lr=self.lr)
         loss_func = nn.MSELoss()
 
-        epoch_iter = tqdm(range(self.max_epoch))
+        epoch_iter = tqdm(range(self.epochs))
         for epoch in epoch_iter:
             opt.zero_grad()
             encoded, decoded = model.forward(input_mat)