Large language model (LLM) agents acting across web applications and online services often need multi-turn corrections to align their work with users' unders…
机构:复旦
来源:arXiv 2611.11560 | AI4Papers 论文推荐平台