DataComp-VLM benchmark launches with over 1,000 controlled experiments to evaluate data curation for vision-language models

🚀New Paper https://arxiv.org/abs/2606.28551

Everyone obsesses over VLM architectures & training recipes. But what about the data?

Presenting the latest work in the DataComp-series: a testbed for VLM data curation with 1,000+ controlled experiments and some surprising lessons 👀

🧵👇