Evaluates the computational runtime of the CellFuse query prediction step.

predict_cells_time(
  dataset_name,
  data_dir,
  test_data_dir,
  test_data,
  model_dir,
  model_date,
  device,
  cluster_column,
  lr,
  margin,
  bs,
  epoch,
  knn_k,
  output_dim,
  dropout_prob,
  activation_function
)

Arguments

dataset_name

Name of the training dataset

data_dir

Path to reference data directory

test_data_dir

Path to query/test data directory

test_data

Name of the test dataset

model_dir

Path to saved model directory

model_date

Date when model was created (this helps CellFuse to load correct model)

device

'cpu' or 'cuda'

cluster_column

Column name containing cluster labels

lr

Learning rate

margin

Margin value for contrastive loss

bs

Batch size

epoch

Number of epochs

knn_k

K for KNN classifier

output_dim

Size of embedding output

dropout_prob

Dropout probability

activation_function

Activation function name

Value

Numeric value indicating runtime (in seconds).

Details

Useful for benchmarking CPU/GPU performance and scalability analysis.