一作OSDI 2026Oral· Operating Systems & Distributed Systems
Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Exploration
Shuzhang Zhong · Haochen Huang · Shengxuan Qiu · Pengfei Zuo · Runsheng Wang · Meng Li
摘要
Tree-of-Thought (ToT) reasoning structures Large Language Model (LLM) inference as a tree-based search, demonstrating strong potential for solving complex mathematical and programming tasks. However, its efficiency is constrained by the _reward dependency barrier_—a synchronization bottleneck cau...