Fetching the paper…

What Happened in LLMs Layers when Trained for Fast vs. Slow Thinking: A Gradient Perspective · Around