The Dataset Used by Microsoft to Enhance Questions and Answers (rajpurkar.github.io)

The Stanford site has been posted on Hacker News previously (links below) and not received much love.

SQuAD2.0 combines the 100,000 questions on 500 wikipedia articles from SQuAD1.1 with over 50,000 unanswerable new questions.

I don't understand why this is effectively ignored, and there are so many folks who discuss GPT-3.

For example: @maraoz about 8 months ago got over 500 responses to https://news.ycombinator.com/item?id=23885684

But less than 10 folks responded to the previous incarnations of this link.

Previous incarnations/links to SQuAD:

@catchmeifyoucan on Dec 6, 2019 https://news.ycombinator.com/item?id=21719929

@davidlee1435 on June 14, 2018 https://news.ycombinator.com/item?id=17308909

@nopinsight on Jan 30, 2018 https://news.ycombinator.com/item?id=16265835

@brittohalloran on Jan 15, 2018 reported it is used by the Alibaba search engine: https://news.ycombinator.com/item?id=16150976

@jonbaer reported on March 1, 2017 https://news.ycombinator.com/item?id=13763816

Well, it has been almost two weeks since I posted this. I guess this shows that many folks interested in Natural Language Processing already know about this site, or don't care to discuss it.