Datasets:

mteb

/

results

like 6

Hello @pengql .
You're supposed to add this metadata to the README.md of your own model. Then, the MTEB leaderboard will automatically fetch it. See e.g. https://huggingface.co/BAAI/bge-large-en-v1.5.
This repository is primarily for external models that e.g. are not on Hugging Face.

• Tom Aarsen

Hello @pengql .
You're supposed to add this metadata to the README.md of your own model. Then, the MTEB leaderboard will automatically fetch it. See e.g. https://huggingface.co/BAAI/bge-large-en-v1.5.
This repository is primarily for external models that e.g. are not on Hugging Face.

• Tom Aarsen

What you mean is that I create an empty model myself, then I add this metadata to readme.md, and then the MTEB ranking list will automatically get it, can you explain it in more detail?

Exactly. However, there are a few more informal requirements: we need the model to be usable, either via an API or by having the weights available. Otherwise your model will be removed from the leaderboard.