Reasoning Ability
Reasoning ability in large language models (LLMs) is a burgeoning research area focused on evaluating and enhancing the capacity of these models to perform multi-step inferences and solve complex problems requiring logical deduction and inductive learning. Current research emphasizes benchmarking LLMs on diverse tasks, including mathematical reasoning, commonsense reasoning, and following procedures, often employing techniques like chain-of-thought prompting and knowledge distillation to improve performance. Understanding and improving LLM reasoning is crucial for building more reliable and trustworthy AI systems with broader applications across various fields, from scientific discovery to decision-making support.
Papers
OPT-R: Exploring the Role of Explanations in Finetuning and Prompting for Reasoning Skills of Large Language Models
Badr AlKhamissi, Siddharth Verma, Ping Yu, Zhijing Jin, Asli Celikyilmaz, Mona Diab
RCOT: Detecting and Rectifying Factual Inconsistency in Reasoning by Reversing Chain-of-Thought
Tianci Xue, Ziqi Wang, Zhenhailong Wang, Chi Han, Pengfei Yu, Heng Ji
Towards Reasoning in Large Language Models: A Survey
Jie Huang, Kevin Chen-Chuan Chang
True Detective: A Deep Abductive Reasoning Benchmark Undoable for GPT-3 and Challenging for GPT-4
Maksym Del, Mark Fishel
Are Deep Neural Networks SMARTer than Second Graders?
Anoop Cherian, Kuan-Chuan Peng, Suhas Lohit, Kevin A. Smith, Joshua B. Tenenbaum
KNIFE: Distilling Reasoning Knowledge From Free-Text Rationales
Aaron Chan, Zhiyuan Zeng, Wyatt Lake, Brihi Joshi, Hanjie Chen, Xiang Ren
Reasoning with Language Model Prompting: A Survey
Shuofei Qiao, Yixin Ou, Ningyu Zhang, Xiang Chen, Yunzhi Yao, Shumin Deng, Chuanqi Tan, Fei Huang, Huajun Chen
Large Language Models are Better Reasoners with Self-Verification
Yixuan Weng, Minjun Zhu, Fei Xia, Bin Li, Shizhu He, Shengping Liu, Bin Sun, Kang Liu, Jun Zhao