Auto-Patching: Enhancing Multi-Hop Reasoning in Language Models

KELMReLMLRMRe-assign community