• 5 Posts
  • 136 Comments
Joined 1 year ago
cake
Cake day: June 25th, 2023

help-circle
  • That is just its core function doing its thing transforming inputs to outputs based on learned pattern matching.

    It may not have been trained on translation explicitly, but it very much has been trained on these are matching stuff via its training material. Since you know what its training set most likely contained… dictionaries. Which is as good as asking it to learn translation. Another stuff most likely in training data: language course books, with matching translated sentences in them. Again well you didnt explicitly tell it to learn to translate, but in practice the training data selection did it for you.




  • Nah. 2k$ was a cheap PR face save for them. Pay 2k$ or deal for weeks and months with “remember how Tesla was a stingy bad corporate and cancelled a large order to a small business without compensation”.

    Noh they can go “Well yeah the cancellation wasn’t exactly gracefully, but hey we compensated the business for it. Our bad.”

    Mind you even just paying the while 15k$ would have been small change for them. So I guess they are not utterly (business relations wise) horrible company, but still a cheap conglomerate.


  • Well difference is you have to know coming to know did the AI produce what you actually wanted.

    Anyone can read the letter and know did the AI hallucinate or actually produce what you wanted.

    On code. It might produce code, that by first try does what you ask. However turns AI hallucinated a bug into the code for some edge or specialty case.

    Hallucinating is not a minor hiccup or minor bug, it is fundamental feature of LLMs. Since it isn’t actually smart. It is a stochastic requrgitator. It doesn’t know what you asked or understand what it is actually doing. It is matching prompt patterns to output. With enough training patterns to match one statistically usually ends up about there. However this is not quaranteed. Thus the main weakness of the system. More good training data makes it more likely it more often produces good results. However for example for business critical stuff, you aren’t interested did it get it about right the 99 other times. It 100% has to get it right, this one time. Since this code goes to a production business deployment.

    I guess one can code comprehensive enough verified testing pattern including all the edge cases and with thay verify the result. However now you have just shifted the job. Instead of programmer programming the programming, you have programmer programming the very very comprehensive testing routines. Which can’t be LLM done, since the whole point is the testing routines are there to check for the inherent unreliability of the LLM output.

    It’s a nice toy for someone wanting to make a quick and dirty test code (maybe) to do thing X. Then try to find out does this actually do what I asked or does it have unforeseen behavior. Since I don’t know what the behavior of the code is designed to be. Since I didn’t write the code. good for toying around and maybe for quick and dirty brainstorming. Not good enough for anything critical, that has to be guaranteed to work with promise of service contract and so on.

    So what the future real big job will be is not prompt engineers, but quality assurance and testing engineers who have to be around to guard against hallucinating LLM/ similar AIs. Prompts can be gotten from anyone, what is harder is finding out did the prompt actually produced what it was supposed to produce.







  • Also not only would they need more satellites, but satellites more densely in any area with multitude of customers. Which eventually hits RF interference saturation.

    Radio signal has only so much bandwidth in certain amount of frequency band. Infact being high up and far away makes it worse. Since more receivers hit the beam of the satellite transmission. One would have to acquire more radio bands, but we’ll unused global satellite transmission bands don’t grow in trees.

    Tighter transmitters and better filtering receivers can help, but usually at great expense and in the end eventually one hits a limit of “can’t cheat laws of physics”



  • However this isn’t about your anecdotal experience. This is about what level of service they can guarantee as minimum and overall to meet the conditions of the subsidy.

    I would also note this isn’t reinstatement matter. FCC refused to give them the subsidy in the first place with this decision. What SpaceX are trying to spin as reneg on previous decision is them making the short list of companies to be considered. Well, getting short listed is not same as being selected fully.

    They passed the criterion for the short list check, but the final authorization and selection included more wide and more through checking on the promises of companies to meet criterion and SpaceX failed the more through final round of scrutiny before being awarded the subsidy.

    Government having awarded bad money previously isn’t fixed by following up bad awards with more bad awards. SpaceX exactly failed since previously money was handed out too losely and FCC has tightened the scrutiny on subsidy awards to not follow up bad money with more bad money.

    Nobody is prevented from buying Starlink, this just means Starlink isn’t getting subsidized with tax payer money.



  • variaatio@sopuli.xyztopics@lemmy.worldBehold
    link
    fedilink
    arrow-up
    10
    ·
    7 months ago

    There possibly is a pushers/braking truck attached to the rear of the Transporter.

    Also one must remember on transporter it is about winning over rolling resistance rather than the weight. Doesn’t necessarily take that powerfull truck on flat ground to pull even great load.

    Also turbine housing has lot of air and as equipment to be lifted to top of a mast, built with light weight in mind. Not for pulling it, but in thought of the crane that has to lift that thing dead load up.



  • Well even on not being rare, lynxes are stalking predators. Given what noisy clumsy travels us humans are and their keen senses, one is lucky to see a lynx. Since firstly they are always stalking or hiding just naturally and specially so hiding upon most likely spotting human way before human spots them. One could go right by one and not notice it. We aren’t on their menu given our size and not being normally encountered prey species. Also as stalkers unless it is something like a mother lynx protecting its young, it won’t make itself known. Far rather hides and let’s you pass without encounter. Since one less risk of the lynx getting injured in fight, if it can’t just hide away and go unnoticed.

    Though on top of that some species of lynx are very endangered.


  • Whatever it is called with that kind of caffeine content you warning label it with listing of exactly how much caffeine it has. Well maybe unless it is named literally “coffee” and is plain brewed coffee and at that brewed coffee with the normal levels of caffeine coffee contains.

    Ones frappe, whippazino also better have needed labels in cases, since given all they mix how the heck one is to know what exactly is the contents. Oh this is extra special “angry frappe” with double squared shot expresso, so exactly how much caffeine is that dear seller per one glass? I just thought you put chili in it or something to make it “angry”, but has literally multiple times more caffeine content.

    This is why all the energy drinks atleast where I live have the ever present “contains high amount of caffeine x mg/100ml”.

    You sell something like that as counter served item with no packaging label to read, well now your menu list must contains at minimum highlights. Something like “our special drunk (HC)” and then somewhere on the menu there reads “HC means high in caffeine”. Then obviously at the counter must be a full labeling booklet of “here is our every product from the plainest brewed coffee to our jumbo mega sandwich and special brew beverage with full nutritional information and ingredients”

    Just like one can’t sell say a pastry in cafe with nut creme filling with out having a big marker on all the menus “contains nuts, nut allergies bevare”. Since similarly nut allergic consuming nuts can be life threatening, well for some people consuming caffeine isn’t healthy and must be disclosed.


  • Specially in say foggy conditions and little bit distance. At which point you won’t clearly maybe differentiate individual elements and more like that’s the rear and “block of light in middle, left and right”. At which point it all little blending one might infact be under impression “the light intensity lowered at the rear, huh, not braking then, did they have they parking break dragging they released or something… ohhhjj shuiiiiiit no it is braking hard”.

    My two cents from here north of Europe and land of snow, rain, fog and occasional white out conditions.


  • Which is the key problem. Everyone is a “responsible gun owner” and “good guy with a gun”… until sometimes they suddenly aren’t anymore. At which point your protection is what was person able to keep under normal circumstances aka what they had in their possession on the moment they had a mental snap.

    Was it a semi-auto shoot as fast as your finger pulls rifle with potentially hundreds of rounds in quick swap magazines or do they have a manual action hunting rifle or shotgun with fixed magazine, that need to be manually reloaded.

    Do they have a pistol with again potentially hundreds of rounds of quick reload ammunition or don’t or maybe a target pistol with fixed magazine.

    That is why places around the world have magazine and type restrictions, since they exactly know “checking backgrounds isn’t fool proof and now amount of background checking helps again sudden newly emerging situation after the checks have been done”.

    Sure that 5 round moose hunting rifle will absolutely wreck say those 5 people, but one can’t exactly run amock shooting around endlessly with moose rifle. Damage limitation. 5 dead people is better situation, than 22 dead people. As cold calculating as that is.



  • variaatio@sopuli.xyztoMastodon@lemmy.ml*Permanently Deleted*
    link
    fedilink
    arrow-up
    19
    arrow-down
    1
    ·
    edit-2
    8 months ago

    Since I is decentralized, it is upto the intances. There is no central authority to eject an instance. Rather other instances individually block the instances they find objectionable to their own criterion.

    At basic its that. Inpractice moderation federations and coalitions etc. have formed among instances of “we maintain joint blocking list and any of us can suggest new additions to it”.

    Due to this one can get ejected from rather sizeable swath, if one one the moderation federations puts one on block list and that is pretty much as far as an “you have 24 hours or we boot you”. You get booted from all the instances part of that federation/coalition.

    Plus stuff like just sources/authors trusted by various instances. “If this guy puts an instance on their published black list, we block. So far that guy has done good job with his list”. Ofcourse instance can at any point decide to not trust that list author anymore.

    So there is no one “how mastodon does it”. Infact this is the one area where “on what instance are you” matters. Since how your home instance decides to do moderation and blocking, that is how your blocking happens. Plus ones personal additions on top.

    Mastodon has a moderation action feature, where one can see listing of what instances and user have been blocked or other moderation action taken. There is explanation field there also for moderator to say “why” but obviously that is upto instance on what their policy is on how exacting their moderation documentation policy is.