LLM-based search agents demonstrate strong capabilities in long-horizon information seeking and reasoning, typically trained via reinforcement learning. Howe…
机构:人大
来源:arXiv 2610.25269 | AI4Papers 论文推荐平台