LLM agents are commonly trained with RLVR, but the sparse nature of typical RLVR rewards makes credit assignment challenging, especially for long-horizon tas…
机构:Meta
来源:arXiv 2610.28388 | AI4Papers 论文推荐平台