Dynamic Sampling that Adapts: Iterative DPO for Self-Aware Mathematical Reasoning

Open in new window