Multimodal language models achieve near-ceiling scores on food recognition benchmarks, yet it remains unclear whether this success reflects genuine cultural …
机构:阿里
来源:arXiv 2609.03526 | AI4Papers 论文推荐平台