Hugging Face and Argilla expand Data Is Better Together with DIBT prompts and multilingual benchmarks

AI Impact Summary

Open-source consortium Hugging Face and Argilla are continuing Data Is Better Together by expanding a shared suite of benchmarks and datasets. Key artifacts include the DIBT/10k_prompts_ranked dataset for prompt ranking, a SPIN-related model ecosystem, and the Multilingual Prompt Evaluation Project (MPEP) with translations across multiple languages. The initiative explicitly addresses language and domain underrepresentation and builds a community around dataset curation, tooling, and cookbook guidance. For technical teams, participating in these datasets and benchmarks will improve cross-language evaluation coverage and provide governance-ready assets to train and evaluate open LLMs.