Fetching the paper…

"Well, Keep Thinking": Enhancing LLM Reasoning with Adaptive Injection Decoding · Around