Add paper link and improve metadata
This PR updates the model card by adding a link to the associated paper and correcting the pipeline tag to feature-extraction, as this model is primarily used for feature extraction. The abstract is also included for more context.
how come this doesn't fall under sentence similarity? most other embedding models are classified as such AFAICT
Both tags are totally fine, but as of late we're noticing a slight shift in favor of feature-extraction over sentence-similarity as the former is more convenient to use over an API. I'm also considering updating the default pipeline tag from sentence-simiarity to feature-extraction.
In the longer term, we'd like to merge the two in some way, but we're not quite sure on that yet.
Also, don't read too much into the PR description from @nielsr . It may or may not be fully AI generated.
sounds good, i'd prefer to keep sentence-similarity for now. i'll happily switch this over to feature extraction once the default pipeline tag is switched over!
Also, don't read too much into the PR description from @nielsr . It may or may not be fully AI generated.
Haha indeed it was LLM generated, but I thought it was doing a good job here :D
I see the model is already linked to the paper page: Training Sparse Mixture Of Experts Text Embedding Models - feel free to claim it with your HF account so it shows up on your profile. You can now also add Github and/or project page URLs to paper pages.