torch_concepts.nn.functional.logic_rule_explanations

logic_rule_explanations(concept_logic_weights: Tensor, concept_names: Dict[int, List[str]] | None = None) → List[Dict[str, Dict[str, str]]]

Extracts rules from rule concept weights as strings.

Parameters:

• concept_logic_weights – Rule embeddings with shape (batch_size, memory_size, n_concepts, n_tasks, 3).

• concept_names – Concept and task names.

Returns:

Rules as strings.

Return type:

List[Dict[str, Dict[str, str]]]