Merge branch 'fix_examples' into revise-gin-dgl-encoder-decoder

autogl/module/model/dgl/_model_registry.py

@@ -22,7 +22,7 @@ class ModelUniversalRegistry:
     @classmethod
     def get_model(cls, name: str) -> _typing.Type[BaseAutoModel]:
         if type(name) != str:
-            raise TypeError
+            raise TypeError(f"Expect model type str, but get {type(name)}.")
         if name not in MODEL_DICT:
-            raise KeyError
+            raise KeyError(f"Do not support {name} model in pyg backend")
         return MODEL_DICT.get(name)

autogl/module/model/pyg/_model_registry.py

@@ -22,7 +22,7 @@ class ModelUniversalRegistry:
     @classmethod
     def get_model(cls, name: str) -> _typing.Type[BaseAutoModel]:
         if type(name) != str:
-            raise TypeError
+            raise TypeError(f"Expect model type str, but get {type(name)}.")
         if name not in MODEL_DICT:
-            raise KeyError
+            raise KeyError(f"Do not support {name} model in pyg backend")
         return MODEL_DICT.get(name)

autogl/module/train/__init__.py

@@ -29,7 +29,12 @@ from .graph_classification_full import GraphClassificationFullTrainer
 from .node_classification_full import NodeClassificationFullTrainer
 from .link_prediction_full import LinkPredictionTrainer
 from .node_classification_het import NodeClassificationHetTrainer
-from .node_classification_trainer import *
+if DependentBackend.is_pyg():
+    from .node_classification_trainer import (
+        NodeClassificationGraphSAINTTrainer,
+        NodeClassificationLayerDependentImportanceSamplingTrainer,
+        NodeClassificationNeighborSamplingTrainer
+    )
 from .evaluation import get_feval, Acc, Auc, Logloss, Mrr, MicroF1
 
 __all__ = [
@@ -42,9 +47,6 @@ __all__ = [
     "GraphClassificationFullTrainer",
     "NodeClassificationFullTrainer",
     "NodeClassificationHetTrainer",
-    "NodeClassificationGraphSAINTTrainer",
-    "NodeClassificationLayerDependentImportanceSamplingTrainer",
-    "NodeClassificationNeighborSamplingTrainer",
     "LinkPredictionTrainer",
     "Acc",
     "Auc",
@@ -53,3 +55,10 @@ __all__ = [
     "MicroF1",
     "get_feval",
 ]
+
+if DependentBackend.is_pyg():
+    __all__.extend([
+        "NodeClassificationGraphSAINTTrainer",
+        "NodeClassificationLayerDependentImportanceSamplingTrainer",
+        "NodeClassificationNeighborSamplingTrainer",
+    ])

autogl/module/train/link_prediction_full.py

@@ -7,8 +7,6 @@ from .evaluation import Auc, EVALUATE_DICT
 from .base import EarlyStopping, BaseLinkPredictionTrainer
 from typing import Union, Tuple
 from copy import deepcopy
-from torch_geometric.utils import negative_sampling
-# from ...datasets.utils import negative_sampling
 from ...utils import get_logger
 
 from ...backend import DependentBackend
@@ -198,6 +196,9 @@ class LinkPredictionTrainer(BaseLinkPredictionTrainer):
             try:
                 neg_edge_index = data.train_neg_edge_index
             except:
+                from torch_geometric.utils import negative_sampling
+                # from ...datasets.utils import negative_sampling
+
                 neg_edge_index = negative_sampling(
                     edge_index=data.train_pos_edge_index,
                     num_nodes=data.num_nodes,

autogl/module/train/node_classification_trainer/__init__.py

@@ -1,14 +1,15 @@
 from ....backend import DependentBackend
 
 if DependentBackend.is_pyg():
-    from .node_classification_sampled_trainer import *
+    from .node_classification_sampled_trainer import (
+        NodeClassificationGraphSAINTTrainer,
+        NodeClassificationLayerDependentImportanceSamplingTrainer,
+        NodeClassificationNeighborSamplingTrainer
+    )
+    __all__ = [
+        "NodeClassificationGraphSAINTTrainer",
+        "NodeClassificationLayerDependentImportanceSamplingTrainer",
+        "NodeClassificationNeighborSamplingTrainer"
+    ]
 else:
-    NodeClassificationGraphSAINTTrainer = None
-    NodeClassificationLayerDependentImportanceSamplingTrainer = None
-    NodeClassificationNeighborSamplingTrainer = None
-    pass
-
-
-
-
-
+    __all__ = []

autogl/solver/classifier/graph_classifier.py

@@ -124,6 +124,8 @@ class AutoGraphClassifier(BaseClassifier):
                     else self._default_trainer[i]
                 )
                 if isinstance(trainer, str):
+                    if trainer not in TRAINER_DICT:
+                        raise KeyError(f"Does not support trainer {trainer}")
                     trainer = TRAINER_DICT[trainer]()
                 if isinstance(model, (tuple, list)):
                     trainer.encoder = model[0]
@@ -219,7 +221,7 @@ class AutoGraphClassifier(BaseClassifier):
 
         set_seed(seed)
 
-        num_classes = max(get_dataset_labels(dataset)) + 1
+        num_classes = get_dataset_labels(dataset).max().item() + 1
 
         if time_limit < 0:
             time_limit = 3600 * 24

autogl/solver/classifier/hetero/node_classifier.py

@@ -115,6 +115,8 @@ class AutoHeteroNodeClassifier(BaseClassifier):
                     else self._default_trainer[i]
                 )
                 if isinstance(trainer, str):
+                    if trainer not in TRAINER_DICT:
+                        raise KeyError(f"Does not support trainer {trainer}")
                     trainer = TRAINER_DICT[trainer]()
                 if isinstance(model, (tuple, list)):
                     trainer.encoder = model[0]

autogl/solver/classifier/link_predictor.py

@@ -134,6 +134,8 @@ class AutoLinkPredictor(BaseClassifier):
                     else self._default_trainer[i]
                 )
                 if isinstance(trainer, str):
+                    if trainer not in TRAINER_DICT:
+                        raise KeyError(f"Does not support trainer {trainer}")
                     trainer = TRAINER_DICT[trainer]()
                 if isinstance(model, (tuple, list)):
                     trainer.encoder = model[0]

autogl/solver/classifier/node_classifier.py

@@ -131,6 +131,8 @@ class AutoNodeClassifier(BaseClassifier):
                     else self._default_trainer[i]
                 )
                 if isinstance(trainer, str):
+                    if trainer not in TRAINER_DICT:
+                        raise KeyError(f"Does not support trainer {trainer}")
                     trainer = TRAINER_DICT[trainer]()
                 if isinstance(model, (tuple, list)):
                     trainer.encoder = model[0]

examples/link_prediction.py

@@ -1,6 +1,8 @@
 from autogl.datasets import build_dataset_from_name
 from autogl.solver.classifier.link_predictor import AutoLinkPredictor
 from autogl.module.train.evaluation import Auc
+from autogl.datasets.utils import split_edges
+from autogl.backend import DependentBackend
 import yaml
 import random
 import torch
@@ -57,6 +59,17 @@ if __name__ == "__main__":
 
     dataset = build_dataset_from_name(args.dataset)
 
+    # split the edges for dataset
+    dataset = split_edges(dataset, 0.8, 0.05)
+
+    # add self-loop
+    if DependentBackend.is_dgl():
+        import dgl
+        # add self loop to 0
+        data = list(dataset[0])
+        data[0] = dgl.add_self_loop(data[0])
+        dataset = [data]
+
     configs = yaml.load(open(args.configs, "r").read(), Loader=yaml.FullLoader)
     configs["hpo"]["name"] = args.hpo
     configs["hpo"]["max_evals"] = args.max_eval
@@ -67,9 +80,7 @@ if __name__ == "__main__":
         dataset,
         time_limit=3600,
         evaluation_method=[Auc],
-        seed=seed,
-        train_split=0.85,
-        val_split=0.05,
+        seed=seed
     )
     autoClassifier.get_leaderboard().show()