scxpand.util.training_utils

Training utility functions for scXpand models.

This module contains common utilities for training operations: - Model validation and setup - Training function routing - Path and configuration validation

Functions

call_training_function(model_type, train_fn, ...)	Call the appropriate training function with standardized parameters.
validate_and_setup_common(model_type[, ...])	Common validation and setup for all main functions.

scxpand.util.training_utils.call_training_function(model_type, train_fn, data_path, save_dir, prm, resume=False, num_workers=0, device=None)

Call the appropriate training function with standardized parameters.

Return type:

dict[str, dict[str, float]]

scxpand.util.training_utils.validate_and_setup_common(model_type, data_path=None, model_path=None)

Common validation and setup for all main functions.

Parameters:

• model_type (ModelType | str) – Type of model to validate

• data_path (str | None (default: None)) – Path to data file (optional, for training/optimization)

• model_path (str | None (default: None)) – Path to model directory (optional, for prediction)

Return type:

tuple[ModelType, ModelSpec]

Returns:

Tuple of (validated_model_type, model_spec)