Why? Almost all of the open source models end up actually sucking for general purpose use, and require a decent bit of effort to even make remotely usable for specialized use cases.
From what I could tell (Corridor Digital mentioned this as well), actual pros use local models through ComfyUI, rather than prompting these proprietary models.
Control and consistency, as well as support for specialized workflows is more important, even if it comes at the expense of model quality.
Because from what I understand, the whole premise of billionaires giving AI all their money is
1. AI succeeds and this success is defined by they will be able to raise prices really high
2. AI fails and receives a bailout
But if there are alternatives I can run locally, they can't raise prices as high because at some point I'd rather run stuff locally even if not state of the art. Like personally, USD 200 a month is already too expensive if you ask me.
not really, in a scenario where demand dies down surplus gpu compute becomes so under-utilized we would see it drop even lower I would think. Prices will go up of course if we keep seeing more tokens needed to solve problems, and demand keeps going up with no increase in efficiency or capacity, which is again not what we're seeing really.
You are saying it needs at most minor tweaks when used for something specific like a product support channel?
For the long-term value of a horribly overpriced and debt ridden corporation it matters quite a lot if the catch up of open source to adequate is less than a decade in instead of several decades in some uses and a few more in others.
There's two scenarios going forward that I can see.
Either there's a massive reduction of inference and training costs due to new discoveries and then those big hardware hoarders have no moat anymore or nothing new is discovered and they won't be able to scale forward.
> ChatGPT has this issue where when it's doesn't know the explanation for something, it often won't hallucinate outright, but create some long-winded confusing word salad that sounds like it could be right but you can't quite tell.
Shellfish is underrated. It has a very convenient tmux integration (auto-restore a specific tmux session per host to work around iOS suspending background apps), supports SSH tunneling via other configured hosts, and can be used as an SFTP file provider for other iOS apps. It’s also generally polished and supports the expected standard terminal features.
There’s a few settings I wished were possible, like using volume buttons as modifier keys in Emacs (I’ve heard about this in other apps), but mostly it works fine.
Blink will end up giving you an experience similar to the stack in doom-coding (as Blink's local capabilities are very limited thanks to iOS rules) except you have to pay a subscription.
Termux on Android will let you do anything you can do on your standard Linux PC.
Hmm, maybe I got grandfathered in or something because I paid some set price a few years ago and have not had a subscription for blink, and just use it the same way I would use Ghostty and then ssh into another machine. Use something else if it needs a sub. Some sibling comments had some recommendations.
Last time I looked at this company they just dumped your uploads into an unauthenticated gcp bucket. They just ran your photo thru an llm and asked for its location at the time, and the founder was doing something very weird (in my opinion) with scraping Tinder profiles.
Back when GeoSpy was available for everyone to use, I did a test where I just uploaded an image that had a black background and white text saying the location of a place and a textual description.
GeoSpy told me that it was the place mentioned in the picture, with the textual description as evidence.
You upload photos to tinder, and tinder has rough data on location provided (distance to you) i believe, the photos were the photos people posted on their tinder profile.
CDK's twin problems are that it compiles down to CloudFormation and that AWS did a terrible job at supporting languages other than TypeScript. The latter is theoretically fixable with a native FFI library that is called from each language, but the former is too leaky of an abstraction.
Considering all the downvotes I got I guess you're not the only one. I'm surprised because I really like cdk. It makes creating an AWS stack really easy, and for having dealt with terraform configurations that were trying to deal with multiple cloud platforms I'd rather have a per-platform eDSL
dumbness usually comes from lack of information, humans are the same way - the difference between other llms is that if opus has information it has a ridiculously high accuracy on tasks.
The pizza report is very annoying to me. It’s a trivial thing for any interested party ( few thousand dollars) to sway up when they want to bring attention to an action the pentagon is doing, and has been such a measure for so long that it’s (in my opinion) ridiculous to still follow it as a source for actual activity. Maybe I should start manipulating it myself.
well, they definitely had an operation tonight to get maduros. maybe the pizza activity is more from the reporters and news crews tgat need to pull a late night shift?
I believe that state actors have put in the relatively low amount of effort to use android device farms to manipulate this statistic. I believe it hasn’t been relevant outside of seeing what the state wants to signal. I also believe individuals can do it as well, since it’s just phones reporting to google maps where they are. You see businesses in tourist cities that are dead empty with massive reviewbots and it showing as a ‘busy place’ as well.
reply