SOURCE-LINKED INTELLIGENCE

DART: Distributional Adversarial Recurrent Training for Algorithm Learning

arXiv · AI, language, vision and robotics · article · Sep 5, 2026 · UTC

Recurrent reasoning models (RRMs) can solve structured problems, achieving easy-to-hard generalization through iterative computation in hidden space. These models are typically trained with instance-level supervision, which becomes increasingly problematic as task difficulty grows: valid solutions occupy a tiny region of the solution space, while invalid solutions proliferate rapidly. We propose Distributional Adversarial Recurrent Training (DART), a training framework that replaces single-point supervision with a local target distribution around the ground-truth solution and aligns model outp

Read original source ↗ Open in workspace

recordType

paper

region

Global