arXiv 2510.24698
ParallelMuse: Agentic Parallel Thinking for Deep Information Seeking
By Baixuan Li, Dingchu Zhang, et al.
Published 2025-10-28
Discussion
Read the public discussion and references gathered around this paper.
Parallel thinking expands exploration breadth, complementing the deep exploration of information-seeking (IS) agents to further enhance problem-solving capability. However, conventional parallel thinking faces two key challenges in this setting: inefficiency from repeatedly rolling out from scratch, and difficulty in integrating long-horizon reasoning trajectories during answer generation, as limited context capacit…