You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi over at MTEB, we would love to make our results more reproducible. One way to do that is to know which model was used to produce a given result (revision id and all).
Would it be possible to make this information accessible from the SentenceTransformers object?
The text was updated successfully, but these errors were encountered:
My understanding is that this should be possible without introducing any additional overhead, as the hf_hub_download helper method caches files based on the repository revision. The repository ID could also be saved when it is successfully used to load a model.
I think it could make sense to incorporate this with #2449, which introduces model_card_data to store some information about the model (which is intended to be used in the model card generation). This already stores base_model, but I can convert that into base_model_id and base_model_revision. Do note that these would be None when loading a local model.
Hi over at MTEB, we would love to make our results more reproducible. One way to do that is to know which model was used to produce a given result (revision id and all).
Would it be possible to make this information accessible from the SentenceTransformers object?
The text was updated successfully, but these errors were encountered: