https://hackerpaper.kaipfister.de/p/2104.08696/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMThttps://news.ycombinator.com/item?id=27008923urn:hackerpapers:mention:2104.08696:27008923Sat, 01 May 2021 19:09:57 GMT<p>Just some days ago researchers from Peking U and Microsoft published a paper[0] saying they can access &quot;knowledge neurons&quot; in pretrained embeddings that will enable &quot;fact editing&quot;[1]. [0]: https://arxiv.org/pdf/2104.08696.pdf [1]: https://medium.com/syncedreview/microsoft-peking-u-researche...</p><p><a href="https://hackerpaper.kaipfister.de/p/2104.08696/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.CLhttps://news.ycombinator.com/item?id=26979416urn:hackerpapers:mention:2104.08696:26979416Thu, 29 Apr 2021 10:31:20 GMT<p>Your first paragraph is spot on. Some people are hell-bent on developing AGI (and these people should exist), and there are some doing SOTA-chasing by tweaking hyperparameters to the extreme and overfitting the data (these works are _mostly_ useless), but most people do not realize the amount of AI we have right now, right this day, is enough bring in fundamental and permanent paradigm changes in as many fields as you can think of. And you are onto something when you say that &quot;empirical records&quot; can be learned by ML models, and a huge amount of grunt work is required for that. Every tom…</p><p><a href="https://hackerpaper.kaipfister.de/p/2104.08696/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.CL