huggingface / datasets

🤗 The largest hub of ready-to-use datasets for ML models with fast, easy-to-use and efficient data manipulation tools
https://huggingface.co/docs/datasets
Apache License 2.0
19.32k stars 2.7k forks source link

Adding ScaNN library to do MIPS? #2124

Open shamanez opened 3 years ago

shamanez commented 3 years ago

@lhoestq Hi I am thinking of adding this new google library to do the MIPS similar to add_faiss_idex. As the paper suggests, it is really fast when it comes to retrieving the nearest neighbors.

https://github.com/google-research/google-research/tree/master/scann

image

lhoestq commented 3 years ago

I haven't played with it (yet) but it sounds really cool !