nirs4all.workspace package

Submodules

• nirs4all.workspace.library_manager module
  • LibraryManager
    • LibraryManager.list_filtered()
    • LibraryManager.list_fullruns()
    • LibraryManager.list_pipelines()
    • LibraryManager.list_templates()
    • LibraryManager.load_template()
    • LibraryManager.save_filtered()
    • LibraryManager.save_fullrun()
    • LibraryManager.save_pipeline_full()
    • LibraryManager.save_template()

Module contents

Workspace management for nirs4all.

Provides library management for saving and loading pipeline templates and trained models.

class nirs4all.workspace.LibraryManager(library_dir: Path)

Bases: object

Manage library of saved pipelines.

list_filtered() → List[Dict]

List all filtered pipelines.

Returns:

List of metadata dictionaries

list_fullruns() → List[Dict]

List all saved full runs.

Returns:

List of metadata dictionaries

list_pipelines() → List[Dict]

List all full pipelines.

Returns:

List of metadata dictionaries

list_templates() → List[Dict]

List all available templates.

Returns:

List of template dictionaries

load_template(name: str) → Dict

Load a template by name.

Parameters:

name – Template name

Returns:

Template dictionary

save_filtered(pipeline_dir: Path, name: str, description: str = '') → Path

Save filtered pipeline (config + metrics only).

Useful for tracking experiments and comparing configurations.

Parameters:

• pipeline_dir – Source pipeline directory

• name – Save name

• description – Optional description

Returns:

Path to saved filtered pipeline

save_fullrun(run_dir: Path, name: str, description: str = '') → Path

Save entire run (all pipelines + binaries + data).

Useful for complete experiment archiving and cross-dataset comparison.

Parameters:

• run_dir – Run directory to save

• name – Save name

• description – Optional description

Returns:

Path to saved run

save_pipeline_full(run_dir: Path, pipeline_dir: Path, name: str, description: str = '') → Path

Save full pipeline (all files + binaries).

Useful for deployment, retraining, and full reproducibility.

Parameters:

• run_dir – Parent run directory (contains _binaries/)

• pipeline_dir – Pipeline directory

• name – Save name

• description – Optional description

Returns:

Path to saved pipeline

save_template(pipeline_config: Dict, name: str, description: str = '') → Path

Save pipeline template (config only, no trained artifacts).

Parameters:

• pipeline_config – Pipeline configuration dictionary

• name – Template name

• description – Optional description

Returns:

Path to saved template file