Hugging Face & Argilla: Building Diverse Datasets - DIBT & MPEP

AI Impact Summary

The Data Is Better Together initiative, a collaboration between Hugging Face and Argilla, has rapidly expanded with community contributions and focused on building diverse datasets. Key achievements include the creation of a 10K-prompt ranked dataset used to train models like SPIN, and the Multilingual Prompt Evaluation Project (MPEP) translating prompts into over 18 languages. This initiative addresses the current lack of comprehensive benchmarks and highlights the need for continued community involvement in building datasets for underrepresented languages and domains.