Haonan Huang, Yuwei Lu, Chengkun Cai · 2026 · arXiv
DOI: 10.48550/arXiv.2610.28066 · open-access PDF
Penilaian multilingual sering mencetak Yes atau No pada pembukaan model, sebuah token yang juga membawa konvensi jawaban model dan bahasa. Kami bertanya ketika cara-cara yang berbeda untuk mendorong dan membaca jawaban mengembalikan penilaian yang sama, dan ketika mereka mengukur kebiasaan jawaban. Kami mengumpulkan 1,1 juta jawaban dari delapan model dalam enam bahasa pada hingga 105 vignettes moral, bervariasi dalam bentuk pertanyaan, pembatasan jawaban, dan membaca jawaban, token atau kalimat, dalam kedua formulir (bateri FAR-6). Profil penilaian tingkat cerita menyatu di seluruh pembacaan dan formulir pertanyaan yang diuji, kecuali yang
Machine translation (Meta M2M100); the English original follows.
Multilingual evaluations often score a model's opening yes or no, a token that also carries model and language answering conventions. We ask when different ways of eliciting and reading an answer recover the same judgment, and when they measure an answer habit. We collect 1.1 million replies from eight models in six languages on up to 105 moral vignettes, varying the question's form, the answer constraint, and the reading of the reply, token or sentence, in both wordings (the FAR-6 battery). Story-level judgment profiles converge across the tested readouts and question forms, except for the ye
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28066, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange