search

MaartenGr / KeyBERT

Minimal keyword extraction with BERT
https://MaartenGr.github.io/KeyBERT/
MIT License
3.31k stars 337 forks source link

How to use leaderboards? #178

Open sdspieg opened 11 months ago

sdspieg commented 11 months ago

Can you recommend a way to find the most appropriate, pretrained recent language model(s) that focuses on semantic similarity AND should work with KeyBERT? E.g. these seem to be appropriate models that should also work for Dutch. But how can we tell which ones would work? Also, we'd like to run KeyBERT with different models, in order to be able to see the differences - you don't happen to have any Jupyter notebooks that would show us how to do this? Thanks!

MaartenGr commented 10 months ago

Personally, I would advise looking at the MTEB Leaderboard. These models are optimized for sentence similarity tasks and especially the top models ("bge-" and "gte-") work incredibly well and can be used within KeyBERT.

sdspieg commented 10 months ago

Great! Thanks much Maarten...