The 2024 Conference on Empirical Methods in Natural Language Processing

From Left:
Professor Mohit Bansal, University of North Carolina at Chapel Hill (PC Chair)
Professor Shafiq Joty, Salesforce Research and Nanyang Technological University (Author)
Jiao Fangkai, Joint PhD student at Nanyang Technological University and A*STAR I2R (Author)
Dr Yaser AI-Onaizan, Saudi Data and AI Authority, National Center for AI (PC Chair)
Professor Yun-Nung (Vivian) Chen, National Taiwan University (PC Chair)

We are proud to announce that the paper, "Learning Planning-based Reasoning by Trajectories Collection and Process Reward Synthesizing," co-authored by Jiao Fangkai (NTU, A*STAR), Qin Chengwei (NTU), Liu Zhengyuan (A*STAR), Nancy F. Chen (A*STAR, NTU), and A/P Shafiq Joty (Salesforce Research, NTU), has been awarded the Outstanding Paper Award at EMNLP 2024.

As one of the most prestigious conferences in natural language processing (NLP), EMNLP highlights innovative research that advances the state of the art in the field. Receiving this recognition reflects the high impact and quality of the team’s work.

The paper tackles key challenges in large language models (LLMs), such as enhancing reasoning reliability and efficiency. By introducing a novel framework that employs Direct Preference Optimization (DPO) with synthesized process rewards, the authors demonstrate solid results, with their model surpassing established benchmarks, including GPT-3.5-Turbo. This research not only advances the capabilities of LLMs but also addresses practical issues like scalability and latency in reasoning tasks.

Details about the paper could be found at ACL Anthology

Official Announcement for Best Papers

Official Recording for the Awarding Ceremony