Bo Ding
2 indexed papers
Publications per year
Top categories
Frequent co-authors
Research Timeline
ParetoPilot introduces a novel zero-surrogate diffusion framework for offline multi-objective optimization, achieving state-of-the-art performance by directly guiding the generation process without relying on external surrogate models.
This paper introduces PASS (Process Advantage Signal Shaping), a method to address three pathologies in GRPO (Group Relative Policy Optimization) for process-supervised reinforcement learning of LLM reasoners.
Papers
Process Advantage Signal Shaping: A Paradigm-Agnostic Middleware for Process-Supervised RL in LLM Reasoners
Chao Wang, Hongtao Tian, Tao Yang, Yunsheng Shi +2 more
This paper introduces PASS (Process Advantage Signal Shaping), a method to address three pathologies in GRPO (Group Relative Policy Optimization) for process-supervised reinforcement learning of LLM r…