data wrappers
Node Classification
- class cogdl.wrappers.data_wrapper.node_classification.ClusterWrapper(dataset, method='metis', batch_size=20, n_cluster=100)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- get_train_dataset()[source]
Return the wrapped dataset for specific usage. For example, return ClusteredDataset in cluster_dw for DDP training.
- class cogdl.wrappers.data_wrapper.node_classification.GraphSAGEDataWrapper(dataset, batch_size: int, sample_size: list)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- get_train_dataset()[source]
Return the wrapped dataset for specific usage. For example, return ClusteredDataset in cluster_dw for DDP training.
- class cogdl.wrappers.data_wrapper.node_classification.M3SDataWrapper(dataset, label_rate, approximate, alpha)[source]
Bases:
cogdl.wrappers.data_wrapper.node_classification.node_classification_dw.FullBatchNodeClfDataWrapper
- class cogdl.wrappers.data_wrapper.node_classification.NetworkEmbeddingDataWrapper(dataset)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.node_classification.FullBatchNodeClfDataWrapper(dataset)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- train_wrapper() cogdl.data.data.Graph [source]
- Returns
DataLoader
cogdl.Graph
list of DataLoader or Graph
Any other data formats other than DataLoader will not be traversed
- class cogdl.wrappers.data_wrapper.node_classification.PPRGoDataWrapper(dataset, topk, alpha=0.2, norm='sym', batch_size=512, eps=0.0001, test_batch_size=- 1)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.node_classification.SAGNDataWrapper(dataset, batch_size, label_nhop, threshold, nhop)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
Graph Classification
- class cogdl.wrappers.data_wrapper.graph_classification.GraphClassificationDataWrapper(dataset, degree_node_features=False, batch_size=32, train_ratio=0.5, test_ratio=0.3)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.graph_classification.GraphEmbeddingDataWrapper(dataset, degree_node_features=False)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.graph_classification.InfoGraphDataWrapper(dataset, degree_node_features=False, batch_size=32, train_ratio=0.5, test_ratio=0.3)[source]
Pretraining
- class cogdl.wrappers.data_wrapper.pretraining.GCCDataWrapper(dataset, batch_size, finetune=False, num_workers=4, rw_hops=256, subgraph_size=128, restart_prob=0.8, positional_embedding_size=32, task='node_classification', freeze=False, pretrain=False, num_samples=0, num_copies=1, aug='rwr', num_neighbors=5, parallel=True)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
Link Prediction
- class cogdl.wrappers.data_wrapper.link_prediction.EmbeddingLinkPredictionDataWrapper(dataset, negative_ratio)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.link_prediction.GNNKGLinkPredictionDataWrapper(dataset)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.link_prediction.GNNLinkPredictionDataWrapper(dataset)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
Heterogeneous
- class cogdl.wrappers.data_wrapper.heterogeneous.HeterogeneousEmbeddingDataWrapper(dataset)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper
- class cogdl.wrappers.data_wrapper.heterogeneous.HeterogeneousGNNDataWrapper(dataset)[source]
Bases:
cogdl.wrappers.data_wrapper.base_data_wrapper.DataWrapper