Recursive large language model (LLM) agents solve long-horizon tasks through decomposition, execution, and environmental feedback. Yet after producing an ato…
机构:中科院
来源:arXiv 2610.32412 | AI4Papers 论文推荐平台