Benchmark the latest `SONAR` models from Meta AI

#24
by reach-vb HF staff - opened

Hi @Muennighoff - Meta AI yesterday released a slew of embedding multimodal embedding models: https://huggingface.co/facebook/SONAR
It'd be great to benchmark the text embedding models.

Massive Text Embedding Benchmark org

Sure, feel free to run them! ๐Ÿ˜‡
I think they use some new tasks, one could add them to the codebase first (https://github.com/embeddings-benchmark/mteb) or just evaluate on the tasks we already have.
Don't hesitate to let me know if you run into any problems!

Sign up or log in to comment