Easy evaluation of the zero-shot capabilities of TinyCLIP

#1
by fhvilshoj - opened

Hi!

Thanks for your big efforts in training this model. It's helping push forward the field of AI!

I wanted to show how we've evaluated the model against a bunch of others on a handful of medical datasets. TinyCLIP does remarkably well on Alzheimers datasets.
It's super easy to add other datasets from HF as well btw.

medical-linear-probe-accuracy.png

We did it with this repo: https://github.com/encord-team/text-to-image-eval

Would be curious to hear how this stacks up against your findings and if it's useful?

Owner

Good job! Thank you for the interest to TinyCLIP : )

Sign up or log in to comment