LogiCoT: Logical Chain-of-Thought Instruction Tuning

Hanmeng Liu, Zhiyang Teng, Leyang Cui, Chaoli Zhang, Qiji Zhou, Yue Zhang

Submission Type: Regular Long Paper

Submission Track: Language Modeling and Analysis of Language Models

Submission Track 2: Theme Track: Large Language Models and the Future of NLP

Keywords: instruction tuning, chain-of-thought, large language model, logical reasoning, GPT-4

Abstract:

Generative Pre-trained Transformer 4 (GPT-4) demonstrates impressive chain-of-thought reasoning ability.
Recent work on self-instruction tuning, such as Alpaca, has focused on enhancing the general proficiency of models. These instructions enable the model to achieve performance comparable to GPT-3.5 on general tasks like open-domain text generation and paraphrasing. However, they fall short of helping the model handle complex reasoning tasks.
To bridge the gap, this paper presents LogiCoT, a new instruction-tuning dataset for Logical Chain-of-Thought reasoning with GPT-4. We elaborate on the process of harvesting instructions for prompting GPT-4 to generate chain-of-thought rationales. LogiCoT serves as an instruction set for teaching models of logical reasoning and elicits general reasoning skills.

Submission Number: 2957