Rubric-based evaluation is widely used to assess LLM-based systems by decomposing response quality into task-specific scoring criteria. However, automaticall…
机构:清华
来源:arXiv 2609.37322 | AI4Papers 论文推荐平台