I can help with "verified correct sources", have a look at "Language Models are Few-Shot Learners" section 2.2 [1]. You may also be interested in Apendix A in

I can help with "verified correct sources", have a look at "Language Models are Few-Shot Learners" section 2.2 [1]. You may also be interested in Apendix A in the same document: "Details of Common Crawl Filtering" [1] https://arxiv.org/pdf/2005.14165.pdf https://arxiv.org/pdf/2005.14165.pdf