Neuro-Symbolic Engine (Fuzzy Logic + LTN/Real Logic Grounding)

What problem does it solve?

The Neuro-Symbolic Engine provides a modular framework to ground symbols into neural modules and evaluate differentiable fuzzy-logic rules, enabling end-to-end neuro-symbolic reasoning with LTN/Real Logic grounding.

Core Features & Use Cases

• Ground symbols to neural predicates and relations, enabling differentiable logic inference over learned representations.
• Support for multi-hop rule networks with attention, symbolic constraint losses, and joint training.
• Use cases include research prototyping for neuro-symbolic AI, knowledge-grounded reasoning in perception pipelines, and open-ended cognitive tasks.

Quick Start

Instantiate the grounding config, register a few predicates and a relation, and run a minimal forward pass to obtain predicate truths and a symbolic constraint loss.