, graph, neg_samples=10, val_split=0.1, data_batch_size=128, graph_batch_size=AUTO, align_burnin=AUTO, safe_burnin=True, max_epochs=AUTO, patience=AUTO, reduce_lr_patience=AUTO, wait_n_lrs=1, directory=None)[源代码]

Fit model on given datasets

  • adatas (typing.Mapping[str, AnnData]) – Datasets (indexed by modality name)

  • graph (Graph) – Guidance graph

  • neg_samples (int) – Number of negative samples for each edge

  • val_split (float) – Validation split

  • data_batch_size (int) – Number of cells in each data minibatch

  • graph_batch_size (int) – Number of edges in each graph minibatch

  • align_burnin (int) – Number of epochs to wait before starting alignment

  • safe_burnin (bool) – Whether to postpone learning rate scheduling and earlystopping until after the burnin stage

  • max_epochs (int) – Maximal number of epochs

  • patience (typing.Optional[int]) – Patience of early stopping

  • reduce_lr_patience (typing.Optional[int]) – Patience to reduce learning rate

  • wait_n_lrs (int) – Wait n learning rate scheduling events before starting early stopping

  • directory (typing.Optional[os.PathLike]) – Directory to store checkpoints and tensorboard logs