About Instruction Tuning Dataset Content

Hi, Authors,

I am very appreciate for your open-sourcing of the OpenELM code and models. I am curious about where I can find the “cleaned variant of UltraFeedback” content. I have also tried checking the paper and Hugging Face, but it does not describe this information in detail.