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

LRM
ReLM
Re-assign community