Answering in the User's Variety: Calibrated Accommodation Pairs for Post-Training

One-line takeaway

CAP enables dialect-aware responses while preserving Standard English, revealing hidden biases in uniform model outputs.

论文提出校准适应配对（CAP）方法，通过偏好数据让模型在保持标准英语的同时更准确匹配用户方言，发现传统训练方式会无意识偏向特定拼写，且统一回复掩盖了训练中的选择性偏见。

Key points

Chat models often ignore user dialects, producing uniform Standard English responses. This paper shows that post-training stages introduce hidden biases favoring certain spellings like British or Australian over AAE or Singlish. The authors propose Calibrated Accommodation Pairs (CAP), a preference data method that rewards dialect matching without compromising standard English for standard users. Experiments across 10 model lines show CAP improves adaptation to non-standard features while preserving instruction-following and avoiding unintended dialect imposition.

• Models favor British/Australian spelling over AAE/Singlish despite user dialect diversity.
• CAP pairs reward dialect density matching while preserving Standard English for standard users.
• CAP improves adaptation to non-standard features without increasing dialect use or harming instruction following.
• Uniform outputs mask selective training biases; audits must examine both output and update process.

Written only from the official arXiv abstract by an open-source model, then checked automatically (every number and name must appear in the abstract). The paper is authoritative.