刘汉蒙（Hanmeng Liu）

1121 School of Computer Science and Technology

58 Renmin Avenue

Haikou 570228

Hainan, P.R. China

I’m a Natural Language Processing researcher at Hainan University.

I obtained my Ph.D. degree at Zhejiang University & Westlake University in 2024.

I was advised by Prof. Yue Zhang at WestlakeNLP Lab. As one of the first members of WestlakeNLP Lab, I designed the badge of the lab, cooperated with Prof Yue Zhang for the NLP MOOC.

After graduation, I joined Huawei as a senior engineer at the department of ICT.

My research interests include large language model reasoning (especially logical reasoning) and the reliability and explainability of large language models in legal, medical, intelligence, and financial domains.

I have a BILIBILI channel. I’m planning to share more contents on LLM tutorials and AI podcasts.

Shoutout to my big bro — I couldn’t have started my PhD journey without him.

news

Mar 31, 2026	I teach Algorithm and Data Structure at BiUH (Bielefeld University of Applied Sciences) as a adjunct professor.
Jan 20, 2026	I give a Keynote at the 2026 AAAI Bridge Program on Logic & AI as a invited speaker and panelist. More information is here. You can watch the recording here.
Apr 12, 2025	I left HUAWEI and joins Hainan University as an associate professor.

latest posts

Sep 21, 2025	当我们在谈论大模型时，我们到底在谈论什么？——一场意外点燃的“智能之火”
Sep 21, 2025	从“做题家”到“造浪者”：华人如何成为全球AI大模型的“隐形统治者”
Apr 15, 2025	大模型约束解码（Constrained decoding）与结构化输出

selected publications

LogiCoT: Logical Chain-of-Thought Instruction Tuning

Hanmeng Liu, Zhiyang Teng, Leyang Cui, and 3 more authors

In Findings of the Association for Computational Linguistics: EMNLP 2023, 2023

Abs Bib HTML PDF Supp Website

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.
@inproceedings{liu2023logicot, title = {LogiCoT: Logical Chain-of-Thought Instruction Tuning}, author = {Liu, Hanmeng and Teng, Zhiyang and Cui, Leyang and Zhang, Chaoli and Zhou, Qiji and Zhang, Yue}, booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2023}, pages = {2908--2921}, year = {2023}, }