affenlehrer

@affenlehrer@feddit.org

• ⁨Comment⁩ on ⁨My friend Charlie is back....⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  Horrible, first time I got one I wondered if it’s a heart attack, aneurysm or if I got hit by a bullet or something…

• ⁨Comment⁩ on ⁨who would win⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  Looong ago I enjoyed watching “the spoony one” on YouTube

• ⁨Comment⁩ on ⁨hot girls hate fascism❤️‍🔥⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  Aren’t they always on the outside?

• ⁨Comment⁩ on ⁨Gotta Philip soon?⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  Look up gas prices in Europe and you’ll feel better. Be aware prices are in liters, not gallons and the Euro is worth more than the dollar

• ⁨Comment⁩ on ⁨It made it interesting when they started with a name and not a number⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  My parents still have a 4 digit number, we also had one until the cable company f’ed up about a year or two ago

• ⁨Comment⁩ on ⁨billy⁩ ⁨⁨2⁩ ⁨weeks⁩ ago⁩:

  Full of coins

• ⁨Comment⁩ on ⁨Important clarification⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  Ok but it also implies that there’s historical data about normal / average amounts of penis licking.

• ⁨Comment⁩ on ⁨Important clarification⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  How does he know?

• ⁨Comment⁩ on ⁨Muad'Dib Dib Dib⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  I read it a few years ago and to me it was a difficult read. Not that I didn’t like it but in my mind it didn’t resemble the movie very much.

• ⁨Comment⁩ on ⁨Nvidia's DLSS 5 Is a Slap in the Face to the Art of Video Game Design⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  I think the idea is that you could use low resolution / detail models that take up less RAM and are faster to process and DLSS hallucinates a high res image.

• ⁨Comment⁩ on ⁨Muad'Dib Dib Dib⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  I like the 1984 up to about 2/3 of it. After that it feels rushed and gets pretty confusing. I also really dislike the shield generator effects.

• ⁨Comment⁩ on ⁨For this you have to give a grade of A+⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  Clearly written by Amy, look at the sad face while thinking

• ⁨Comment⁩ on ⁨463 games, probably only played 1/10th of them⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  Humble bundle is the reason for this…

• ⁨Comment⁩ on ⁨A Fallout 4 QA tester nuked the RPG so hard that Zenimax executives got emails about it: "I was running around super-nuking the entire wasteland and found 4 crashes in a single morning"⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  I also did that (and some development) and hated that basically every phone behaved differently, even sibling models from the same manufacturer. J2ME was a nice idea but every phone had different glitches. The emulators were shit. So we had almost all models of all major brands there and had to test everything on all of them.

• ⁨Comment⁩ on ⁨Lik the bone⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  I remember as a child when I was digging in the sandbox I dug up a hole lot of cat poop and “wet” sand. Since I’m still alive today it probably wasn’t too bad but it doesn’t really feel like a good idea.

• ⁨Comment⁩ on ⁨Lik the bone⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  Isn’t that a health risk?

• ⁨Comment⁩ on ⁨Sometimes I wonder if the subtext of religion is; if you join one youre telling on yourself that you can easily be manipulated.⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  I think my phrasing wasn’t very good. Creating positive experiences for oneself makes sense regardless of life after death.

  However, some behavior is motivated by long term consequences. These matter less or differently if you believe there is no afterlife.

  E.g. some people behave differently in at a party with people they will never see again than a party with e.g. colleagues and family.

  Or many people leave public toilets dirty while they keep the one at home clean.

  People who believe in an afterlife might have the mindset they will see people from their life again and they might (if they believe in reincarnation) walk the earth again. So it’s a motivation to keep the environment liveable for future generations (beyond just family and kids) or to be nice to people you likely don’t meet again.

  The behavior could be identical for people with or without a belief in an afterlife but their motivation and thought process behind it might be different.

• ⁨Comment⁩ on ⁨Rude⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  Superior assholes

• ⁨Comment⁩ on ⁨Sometimes I wonder if the subtext of religion is; if you join one youre telling on yourself that you can easily be manipulated.⁩ ⁨⁨3⁩ ⁨weeks⁩ ago⁩:

  Sounds like more like an emotional response than not caring to me.

• ⁨Comment⁩ on ⁨Sometimes I wonder if the subtext of religion is; if you join one youre telling on yourself that you can easily be manipulated.⁩ ⁨⁨4⁩ ⁨weeks⁩ ago⁩:

  It’s good that you don’t give a fuck but not everyone is able to be that way. So if you give fucks and a religion makes you feel better, why not?

• ⁨Comment⁩ on ⁨Sometimes I wonder if the subtext of religion is; if you join one youre telling on yourself that you can easily be manipulated.⁩ ⁨⁨4⁩ ⁨weeks⁩ ago⁩:

  But that’s only if you believe in non-existence after death. The thought that ones choices are inconsequential might (in some people) also lead to harmful behavior (eg harming later generations with environmental destruction etc).

  However, whatever one believes it shouldn’t be due to doctrine, it should be the result of an active though process.

• ⁨Comment⁩ on ⁨Sometimes I wonder if the subtext of religion is; if you join one youre telling on yourself that you can easily be manipulated.⁩ ⁨⁨4⁩ ⁨weeks⁩ ago⁩:

  I guess that’s true for many. However, there is also a possibility to face it and then “handle it” by taking Pascal’s wager.

  In my opinion non-existence is in the center of the fear. If you stop existing after death whatever you decide to do now doesn’t matter any more. If you believe in existence after death (part of most religions) and it makes you happy, why not?

  True that there are extremist sects of almost any religions and religion is a common cause of fights but I think that’s more rare nowadays and many mainstream religious organizations preach peace and tolerance.

• ⁨Comment⁩ on ⁨Sometimes I wonder if the subtext of religion is; if you join one youre telling on yourself that you can easily be manipulated.⁩ ⁨⁨4⁩ ⁨weeks⁩ ago⁩:

  That’s a legitimate fear IMO.

• ⁨Comment⁩ on ⁨spoopy figs⁩ ⁨⁨4⁩ ⁨weeks⁩ ago⁩:

  Ethical vegans want to avoid suffering. If figs cause suffering is a philosophical question.

• ⁨Comment⁩ on ⁨How to I prove to someone that the U.S. moon landing wasn't staged?⁩ ⁨⁨4⁩ ⁨weeks⁩ ago⁩:

  The flat earth people often say it’s a hologram

• ⁨Comment⁩ on ⁨The Hole⁩ ⁨⁨5⁩ ⁨weeks⁩ ago⁩:

  Dangerous

• ⁨Comment⁩ on ⁨ard⁩ ⁨⁨5⁩ ⁨weeks⁩ ago⁩:

  Doesn’t work in German (also a Germanic language like English):

  • Mustqerd: Senf / Mosterd
  • Bastard: Bastard
  • Wizard: Zauberer
  • Coward: Feigling
  • Drunkard: Säufer / Trunkenbold
• ⁨Comment⁩ on ⁨don't 👏 claim 👏 to 👏 be 👏 a 👏 fan 👏 of 👏 tag 👏 team 👏 wrestling 👏 if 👏 you 👏 don't 👏 know 👏 who 👏 this 👏 is⁩ ⁨⁨5⁩ ⁨weeks⁩ ago⁩:

  Ah, the Duras sisters…

• ⁨Comment⁩ on ⁨ELI5. Limit of current gen AI/LLMs⁩ ⁨⁨5⁩ ⁨weeks⁩ ago⁩:

  I have mixed feelings about it. I wouldn’t give code a full production application but I think it’s sometimes helpful if the LLM is able to generate a prototype or scaffold to get a head start. Removes some of the friction of starting a project.

  The fully vibe coded stuff I’ve seen so far were usually unmaintainable dumpster fires.

• ⁨Comment⁩ on ⁨ELI5. Limit of current gen AI/LLMs⁩ ⁨⁨5⁩ ⁨weeks⁩ ago⁩:

  The LLMs will just predict probabilities for the single next token based on all previous tokens in the context window (it’s own and the ones entered by the user, system prompt or tool calls). The inference engine / runtime decides which token will be selected, usually one with high probably but that’s configurable.

  The LLM can also generate (predict) special tokens like “end of imaginary dialogue” to end it’s turn (the runtime will give the user a chance to reply) or to call tools (the runtime will call the tool and add the result to the context window).

  Some of the tool calls are e.g. web searches etc. and the search results will be added to the context window. The LLM can decide to do more calls for further research, save data in “memory” that can be accessed by later “sessions”.

  Models tend to get larger context windows with every update (right now it’s usually between 250K - 2M tokens but the model performance usually gets worse with more filled context windows (needle in a hay stack).

  To keep the window small agentic tools often “compact” the context window by summarizing it and then starting a new session with the compacted context.

  Sometimes a task is split into multiple sessions (agents) that each have their own context window. E.g. one extra session for a long context subtask like analysis of a long document with a specific task and the result is then sent to an orchestrator agent in charge of the big picture.