Case Study: Hugging Face improves AI model discoverability with Meilisearch

Hugging Face facilitates AI accessibility with Meilisearch

Hugging Face, an open-source provider of machine learning technologies, needed a more flexible and typo-tolerant full-text search solution for its platform. Their previous system was insufficient for helping users discover over 2 million AI models, datasets, and demos based on the rich metadata in their model cards. They chose to implement Meilisearch Cloud to address this challenge with discoverability.

The solution from Meilisearch provided superior customizability for ranking rules and offered out-of-the-box relevancy. As a result, Meilisearch now powers the discovery of 2.2 million model cards, 500,000 datasets, and 60,000 demos on the Hugging Face Hub. This implementation gives users a choice between keyword filtering and a powerful full-text search that scans the entire content of model cards, greatly enhancing the user experience and supporting AI democratization.

Hugging Face

Mishig Davaadorj

Software Engineer