Key ideas:
- Learn structural patterns through similarity, not gradient descent - Knowledge reacts to being learned and forms relationships with other knowledge - Hierarchical pattern matching works with nested structures - Event-driven responses for compositional reasoning - Fully interpretable - every decision traceable to learned patterns
Built this after months researching abstract reasoning (ARC challenge) and interpretable systems. Traditional approaches felt too rigid - this uses active, self-organizing knowledge as primitives. Early stage but core works. Applications: ARC, interpretable AI, few-shot learning, autonomous agents. pip install general-intelligence
Show HN: General Intelligence – Active knowledge framework for machine learning
github.com