Another model - I don't quite remember which, I think it was one of GPT ones? - didn't have access to thinking traces after it finished the thought - they simply were removed from the context to save tokens. Can it be the same with Gemini? Maybe it just doesn't know what it did/thought in the previous turn and so it hallucinates that it doesn't have context access function.

No, it works just fine? And this particular problem i think is date pre-elon because people were complaining about disappearing likes for a few years.

Maybe through the app, but I get sent links to x things all the time and they just straight up don't load. Same thing with links to old x threads linked from different websites.

Surprisingly it does show these posts on your "Likes" page in your profile. But they aren't marked as liked, and liking one will increase the like indicator.

> But they aren't marked as liked, and liking one will increase the like indicator.

https://en.wikipedia.org/wiki/Eventual_consistency

In addition to "optimistic UI". https://imhoff.blog/posts/optimistic-ui-primer

The UI optimistically allows the user to tap/click the like button and the eventually-consistent backend determines if the like is valid. The UI doesn't verify if the user is allowed to like because that's the backend's job.

Tom Scott has a video on it. https://youtube.com/watch?v=RY_2gElt3SA

this explains so much, thank you!

every day a dozen critical of Musk/DOGE discussions land on HN main, and zero non-critical/positive ones

it is an illusion of being surrounded by brigades of conservatives censoring you, if anything - the opposite is true

when I tried submitting a positive take on recent happenings in DC - my own post was flagged withing 5 minutes, and I was chastised for posting a link to nazi platform (it was a link to X longpost)

It is a nazi platform, in that it's owned and controlled by a nazi.

Please control yourself and stop being hysterical

I note that you didn't refute what I said.

Despite the recent deluge of criticism on HN, this perspective offers important counterpoints about the new administration's actions.

So, with many absolutely random people (including me) getting DALL-E2 invites past few days this link to prompt engineering guide should help us create better art. Probably will also work with Midjourney which also opened beta access to everyone yesterday

Sadly, is seems you can't get on Dev or Beta channel since June 24th (if you weren't enrolled before), so you won't be able to get the Insider build with WSLg on your existing win10. Only if you install it from clean ISO and update. Source: https://answers.microsoft.com/en-us/insider/forum/all/how-in...

I tried it now and you could only choose Release Preview channel which still has only build 19044.

Russian language source: https://zona.media/news/2021/10/04/otb-1

"Gulagu.net (Russian human rights organization dedicated to the protection of prisoners' rights) received a video from a former prisoner - programmer, who "was beaten and tortured, and then prison's administration decided to use him in his professional capacity"

The two books I usually recommend for HPMOR fans are Worm (https://parahumans.wordpress.com/) and The Mother Of Learning (https://www.royalroad.com/fiction/21220/mother-of-learning) Both considered to be a part of "rational fiction" sub-genre the same as HPMoR.

I have two phones in the house that are affected - Huawei P10 and Xiaomi Mi Note 3

Huawei simply doesn't let me use google app, but xiaomi has become useless - it has annoying pop-up asking for crash-report every ten seconds.

Rolling back updates helped for a couple of minutes, after that crashes started again.