TypeError: _load_sbert_model() received unexpected keyword argument 'token' when initializing INSTRUCTOR model in Colab Notebook

[devin-liu]

Running into a TypeError while trying to load the INSTRUCTOR model from the InstructorEmbedding package in a Google Colab notebook. The error message says _load_sbert_model() got an unexpected keyword argument 'token' but I am unaware of how that 'token' argument is getting activated or passed.

[Adebesin-Aramide]

Hi, did you figure out the issue?
I'm stuck here as well

[anhalu]

I got the same issues, any updated ? I understand that the library only works with SentenceTransformer version 2.2.2 but that version is too old for now. pls update lib version.

[JosephGatto]

For those also facing this issue, it seems like this still works: https://sbert.net/docs/sentence_transformer/pretrained_models.html#instructor-models

[devin-liu]

Hey everyone, I managed to get past this step, but I can’t recall the exact way I did it. Here’s the notebook link if someone could run the same commands in the cells I used and report back what was missing in the instructions:

https://colab.research.google.com/gist/devin-liu/be0c89afa78a2750a9fec2ea8e654524/instructor-embedding.ipynb#scrollTo=QFszaK3SsM_F

[BBC-Esq]

Please use the "instructor.py" script in my fork of this repository and it should hopefully resolve all issues. It's compatible with the most recent version of sentence transformers as of the date I'm posting this. Here's the repo:

https://github.com/BBC-Esq/instructor-embedding

Unless/until the repository owner accepts my pull request, even if you pip install you'll still have to manually replace instructor.py with the version in my repo, just FYI.