• 19 Posts
  • 847 Comments
Joined 5 years ago
cake
Cake day: December 20th, 2021

help-circle










  • Under new rules rolling out over the coming months, a small number of users will be required to leave some of their moderator posts so that they aren’t moderating more than five subreddits with 100,000 monthly visitors.

    Good. We could do with a similar rule here on lemmy. I know there’s probably a lack of people to fill all the needed moderator positions, but I’m pretty sick of seeing the same people moderating the majority of communities that I visit.

    It becomes especially apparent when someone gets banned from one community and the powermod bans them from every other community that they moderate. It’s rare that someone actually deserves that type of scorched-earth response.











  • The problem is that an LLM is a language model, not an objective reality model, so the best it can do is estimate the probability of a particular sentence appearing in the language, but not the probability that the sentence represents a true statement according to our objective reality.

    They seem to think that they can use these confidence measures to filter the output when it is not confident of being correct, but there are an infinite number of highly probable sentences in a language which are false in reality. An LLM has no way of distinguishing between unlikely and false, or between likely and true.