scglue.models.scglue.SCGLUEModel.fit¶
-
SCGLUEModel.
fit
(adatas, graph, edge_weight='weight', edge_sign='sign', neg_samples=10, val_split=0.1, data_batch_size=128, graph_batch_size=- 1, align_burnin=- 1, safe_burnin=True, max_epochs=- 1, patience=- 1, reduce_lr_patience=- 1, wait_n_lrs=1, directory=None)[source]¶ Fit model on given datasets
- Parameters
adatas (
Mapping
[str
,AnnData
]) – Datasets (indexed by domain name)graph (
Graph
) – Prior graphedge_weight (
str
) – Key of edge attribute for edge weightedge_sign (
str
) – Key of edge attribute for edge signneg_samples (
int
) – Number of negative samples for each edgeval_split (
float
) – Validation splitdata_batch_size (
int
) – Number of cells in each data minibatchgraph_batch_size (
int
) – Number of edges in each graph minibatchalign_burnin (
int
) – Number of epochs to wait before starting alignmentsafe_burnin (
bool
) – Whether to postpone learning rate scheduling and earlystopping until after the burnin stagemax_epochs (
int
) – Maximal number of epochsreduce_lr_patience (
Optional
[int
]) – Patience to reduce learning ratewait_n_lrs (
int
) – Wait n learning rate scheduling events before starting early stoppingdirectory (
Optional
[PathLike
]) – Directory to store checkpoints and tensorboard logs
- Return type