Hugging Face「Data Is Better Together」社群數據協作計劃：回顧與展望

Original: Data Is Better Together: A Look Back and Forward

### Background In the current development of large language models (LLMs), high-quality alignment data (such as the preference data…

Hugging Face 發表「Data Is Better Together (DIBT)」計劃的階段性回顧與展望。該計劃旨在透過社群協作，解決高品質開源對齊數據不足的痛點。過去一年中，DIBT 透過 Argilla 等工具成功匯聚全球開發者，共同標註與生成了多個高質量的偏好與提示詞數據集。未來，該計劃將進一步擴展任務類型，推動更具包容性與去中心化的 AI 數據生態。

Summaries are AI-generated; the original article is authoritative.