Multilingual evaluation benchmarks are the primary instrument for assessing whether large language models generalize beyond English, yet the adequacy of thes…
机构:Microsoft
来源:arXiv 2610.36990 | AI4Papers 论文推荐平台