logoalt Hacker News

tonypapousektoday at 2:48 AM1 replyview on HN

It’s a shame of bulk of that training data is likely 2010s blogspam that was poor quality to begin with.


Replies

2god3today at 2:56 AM

But isn't that a reflection of reality?

If you've made a significant investment in human capital, you're even more likely to protect it now and prevent posting valuable stuff on the web.