Thank you for 1 million downloads on PyPi 🎉

[xhluca]

Really glad to see that this library is useful to many!

Would love to hear what you built with BM25S in this discussion!

[BUAADreamer]

Our RAG Project EasyRAG adopts bm25s to implement bm25 search. It speeds up 50x than official implementation! Thanks for your excellent work!!!

[xhluca]

Glad to know it helps power cutting edge RAG projects!

[tomaarsen]

Congratulations! The first million is the hardest 😄
I've used your implementation as recently as ~3 days ago when I evaluated BM25 for https://huggingface.co/blog/static-embeddings! I already had it installed though, so I'm afraid I'm not a part of the recent growth.

• Tom Aarsen

[xhluca]

Your new static embedding project looks great! It is so fast too, and seems like it can be scaled with FAISS too

[tomaarsen]

Should indeed work like a charm!

[jacoobes]

We had used it in development in MikeGPT, an AI chatbot tasked to help students and faculty for Louisiana State University. We have since moved away from it but had been an amazing tool to use. Highly recommend!

[xhluca]

Good to see the impact for the education domain! Is MikeGPT open source?

[jacoobes]

It is not, however I am eternally grateful for the open source community, being an open source contributor as well!

[Topping1]

I had used it to build BM25 PDF Search, a research application for large PDF collections (similar to Semantra). Your library really makes a difference in search response times for large (10 GB+) PDF collections!. Thanks for your work!.

[xhluca]

Glad it was able to help you!