作者位次 #4/10ACL 2026Poster· Natural Language Processing
R^3 : End-to-End Reasoning-based Planning for Multi-step Retrosynthesis via Reinforcement Learning
YiFei Wang · Qizhi Pei · Jiangtao Feng · Yuntian Shi · Yi Duan · Lihao Wang · LEI BAI · Lijun Wu · Wei-Ying Ma · Hao Zhou
摘要
Multi-step retrosynthetic planning is a fundamental challenge in organic chemistry, traditionally modeled as a combinatorial search problem guided by single-step prediction models. However, this search-centric paradigm often disconnects from the explicit chemical reasoning processes employed by h...