Interactive Reasoning: Visualizing and Controlling Chain-of-Thought Reasoning in Large Language Models