Listen to this story
On Thursday, Mishig Davaadorj from Hugging Face, announced the launch of their full-text search engine, hf.co/search. The search allows users to do a full-text search from over 200k models, datasets, and spaces. This includes all LLM models, graphs, and others hosted on the website.
The search engine, as demonstrated in the video, now also is highly flexible, and is no longer limited to exact string matches, and can overcome spelling mistakes by doing text search, instead of fuzzy search matching. Users can also share their search queries by copying the link. The search engine also supports dark mode.
With the text matching feature, finding models and datasets on Hugging Face is now a lot easier. Compared to the previous search, by searching a single word, users now get all the libraries that have that keyword.
Julien Chaumond, CTO of Hugging Face, said in his LinkedIn post that they are soon going to release a blog with the documentation to explain the tech behind it.
Just recently, GitHub announced the launch of their new code search engine, BlackBird that was built on Rust. The search engine maintains a code search index shared by Git blob object ID and enables accessing over 45 million GitHub repositories. It is being described as the game changer as the default search of GitHub as it does not support special characters search.
The search engine upgradation trend has been going on since Google and Bing are also looking to upgrade their systems with AI and chatbots.