Blazing Fast SetFit Inference with 🤗 Optimum Intel on Xeon

[tomaarsen]

Hello!

Pull Request overview

• Add "Blazing Fast SetFit Inference on Intel Xeon" blogpost

Details

This blogpost follows the notebook contributed by Intel showcasing the performance optimizations possible for SetFit models using optimum-intel.

TODOs:

• [x] Twice we refer to the "notebook", which is currently https://github.com/danielkorat/setfit/blob/faster-optimum-intel-nb/notebooks/setfit-optimum-intel.ipynb. After https://github.com/huggingface/setfit/pull/489 is merged, we should link to https://github.com/huggingface/setfit/blob/main/notebooks/setfit-optimum-intel.ipynb instead.
• [ ] Merge https://github.com/huggingface/setfit/pull/489 (await final changes by @danielkorat)

@echarlaix Feel free to assist me with reviewing. cc @danielkorat @MosheWasserb @orenpereg

• Tom Aarsen

[tomaarsen]

Note for reviewers: The blogpost is still being edited in a Google Docs environment outside of this PR, so expect fairly notable changes to be merged into this PR later.

[tomaarsen]

Thanks for the review @echarlaix! Your suggestions look very sensible.

@danielkorat @orenpereg @moshewasserb What is your ideal release date?

[orenpereg]

Hi @tomaarsen, we are just waiting for Intel approval to publish performance numbers on Intel CPU. hope to get it in the next few days.

[tomaarsen]

All good! Thanks for the heads up.