visualize_random_samples

hybrid_learning.experimentation.fuzzy_exp.fuzzy_exp_vis.visualize_random_samples(experiment_root, model_key, split, logic_type, formula_dirs=None, changed_constants=None, num_samples=15, only_filenames=None, skip_missing='warn', save_as=None, additional_formula_mods=None, verbose=False)

For the first num_samples samples visualize the cached concept and recalculated formula output masks. The formula objects given by additional_formula_mods are recalculated based on the given changed_constants, and displayed together with the concept outputs. The original formula objects and settings are loaded from the sacred experiment logs, the concept outputs are loaded from cache (defined in experiment settings). For the experiment logs, the following folder structure is assumed: experiment_root/model_key/split/<formula_dir>/<logic_type>/logs/

Parameters

• additional_formula_mods (Optional[Dict[str, Callable[[Merge, Merge], Merge]]]) – see ResultsIterator

• changed_constants – see ResultsIterator

• num_samples (int) – number of samples to plot

• skip_missing (Union[bool, str]) – see to_logic_dirs()

• save_as (Optional[str]) – see show_and_save_plot()

• experiment_root (str) –

• model_key (str) –

• split (str) –

• logic_type (str) –

• formula_dirs (Optional[list]) –

• only_filenames (Optional[Iterable]) –

• verbose (bool) –

Return type

List[Dict[str, Any]]