Fetching the paper…

How Well do LLMs Compress Their Own Chain-of-Thought? A Token Complexity Approach · Around