Unstructured-IO
feat: Add Mixedbread AI Integration
Hey guys! This is from Mixedbread AI. With mixedbread we're building SOTA models and tools to streamline retrieval. So far, we trained some of the most widely used open-source embeddings and reranking models (https://huggingface.co/mixedbread-ai).
We notice that Unstructured is providing embeddings integrations. We would love to partner up and contribute to this project. @MthwRobinson
Thanks for the contribution @huangrpablo ! We'll review this as soon as we're able.
Looks like potentially some dependency conflicts. Running make pip-compile from a Python 3.9 environment will likely fix that.
- https://github.com/Unstructured-IO/unstructured/actions/runs/9911978833/job/27385814239?pr=3392
From the CI on the "clone PR", you may also need to update the version in unstructured/__version__.py (see this job)
@MthwRobinson just made the fixes. make pip-compile also resulted in the dependency changes of other connectors.
@MthwRobinson hey, I put the dependencies of other integrations back to untouched. Could you have a look and run the CI again if it looks fine? Thanks!
@MthwRobinson Hey, would love to hear any update on this!
@huangrpablo I'll take over for Robinson from here. I'm gonna check it out today.
copying this over to https://github.com/Unstructured-IO/unstructured/pull/3513 so that its easier to run CI, etc.