Guangyi Liu

Email: guangyiliu.xx [AT] gmail.com

I currently serve as a Postdoctoral Fellow at MBZUAI/CMU, where I have the privilege of working with Prof. Eric Xing and Prof. Zhiting Hu. Before, I studied at CUHK-Shenzhen and obtained my Ph.D. degree from CUHK in 2023. Additionally, I obtained my B.Eng. degree from UESTC in 2018.

My research interests mainly lie in general areas in text generation, generative models, and large pretrained language model. Collaborations and discussions on related topics are welcome.

Apart from my academic pursuits, I have a passion for Powerlifting. To give you a glimpse of my milestones: I've achieved a Squat of 196kg, Bench Press of 120kg, and Deadlift of 204kg, all at a body weight of 80kg.

Google Scholar  /  Github  /  LinkedIn  /  Our Lab

• January 2024: Invited to serve as Reviewer for ACL ARR, and COLM.
• December 2023: Invited to serve as Reviewer for ICML 2024.
• November 2023: Invited to give a talk to the Open Innovation AI Research Community Workshop host by Meta AI in London. [slides]
• October 2023: Invited to serve as Reviewer for ACM Computing Surveys (IF=16.6).
• August 2023: Invited to serve as Reviewer for ICLR 2024.
• June 2023: Commenced my role as a Postdoctoral Fellow.
• May 2023: Invited to serve as PC Member for IJCNLP-AACL 2023.
• April 2023: Invited to serve as a Reviewer for IEEE Transactions on Medical Imaging.

• March 2023: Invited to serve as a Reviewer for NeurIPS 2023.
• December 2022: I gave a series of presentations about very details of Transformer model at CIAI center. And we had lots of insightful discussions! [slides]
• November 2022: Invited talks at King's College London and University of Warwick: "Composable Text Controls in Latent Space with ODEs". [slides]
• September 2022: Start visiting at MBZUAI in the UAE.
• August 2022: Our preprint paper Composable Text Controls in Latent Space with ODEs has been released on arXiv.
• May 2022: Our paper Don't Take It Literally: An Edit-Invariant Sequence Loss for Text Generation has been accepted by NAACL 2022 as an oral presentation.
• July 2021: Our paper Medical-VLBERT: Medical Visual Language BERT for COVID-19 CT Report Generation with Alternate Learning has been accepted by TNNLS 2021.
• October 2020: Our paper Learning to Decouple Relations: Few-Shot Relation Classification with Entity-Guided Attention and Confusion-Aware Training has been accepted by COLING 2020.