Large Language Models Have Intrinsic Meta-Cognition, but Need a Good Lens

LLMAGLRMRe-assign community