Self-Training with Direct Preference Optimization Improves Chain-of-Thought Reasoning

Open in new window