Full set of 38-DataComp results for ViTamin models

Hey, thanks so much for releasing such great models!

I was wondering if it would be possible to release the full set of per-dataset results for the 38 downstream datasets in datacomp-evals. Currently, the paper only reports the average 38 dataset performance number, but I am interesting in getting each of those numbers independently. 

Thanks in advance, I would greatly appreciate your help!