Reasoning Elicitation in Language Models via Counterfactual Feedback

Alihan Hüyük, Xinnuo Xu, Jacqueline Maasch +2

Despite the increasing effectiveness of language models, their reasoning capabilities remain underdeveloped. In particular, causal reasoning through counterfactual question answeri…