Recursive self-improvement (RSI) of agent harnesses has emerged as an effective approach to enhancing the capabilities of large language model (LLM) agents. …
机构:清华
来源:arXiv 2610.28201 | AI4Papers 论文推荐平台