A reasoning strategy where a model explores multiple candidate reasoning paths as branches, evaluates them, and can backtrack from weak branches to try others.
Continue to AI University →