Runcong Zhao

Steer LLM reasoning using a single continuous token, without relying on SFT or RL. Check out our #ICML2025 poster E-2024 from the KCL NLP group: Soft Reasoning: Navigating Solution Spaces in Large Language Models through Controlled Embedding Exploration. Paper: arxiv.org/abs/2505.24688 GitHub: github.com/alickzhu/Soft-…

We developed OpenToM, a benchmarking dataset for evaluating LLMs’ Theory-of-Mind capabilities. My PhD student Hainiu Xu @seacow_x is presenting our work at #ACL2024. @kclnlp @kclinformatics Paper: arxiv.org/abs/2402.06044 Dataset: seacowx.github.io/projects/opent…

KCL NLP researcher Runcong and Qinglin is presenting their work on character relationship extraction from detective novels at board 49. @runcongz @alickzhu79 @yulanhe

Members of our lab will be presenting their works at @aclmeeting next week in Bangkok. Come and check them out if you are around. We would love to hear your thoughts and advices. #ACL2024 @kclinformatics, @yulanhe, @zhengyuan_nlp, @HYannakoudakis, Lin Gui, Oana Cocarascu

Introducing NarrativePlay: a novel system that allows users to role-play a fictional character and interact with other characters in narratives in an immersive environment. Join us for an immersive experience at the #AAAI2024 demo session, 6:15 PM today.

Ever wondering how well can LLMs read people’s mind? Introducing OpenToM, a comprehensive benchmarks for evaluating LLMs’ Theory-of-Mind capabilities…… (story in figure is simplified, check the dataset for actual OpenToM story!)

I will be presenting our #SIGIR2023 paper "Cone: Unsupervised Contrastive Opinion Extraction" tomorrow at 11 am. Join us at @SIGIRConf in Session 21. @yulanhe arxiv.org/abs/2305.04599

Excited to share our #ACL2023NLP paper which proposes a novel dynamic Brand-Topic Model, tracking brand scores and polarity-bearing topics over time. Join us at Poster Session 1. @yan_hanqi @yulanhe #NLProc #sentimentanalysis #TopicModel arxiv.org/abs/2301.07183.

Come to see us at #AIUK to talk and try PANACEA: an AI-empowered veracity assessment system on COVID-19 related claims. panacea2020.github.io