Fetching the paper…

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