Would it be that shocking if they had bots here?

Well there are 2 things.

First there is speed for which they do indeed rely on multiple thousands of super high end industrial Nvidia gpus. And since the 10Billion investment from microsoft they likely expanded that capacity. I’ve read somewhere that chatgpt costs about 700,000 a day to keep running.

There are a few others tricks and caveats here though. Like decreasing the quality of the output when there is high load.

For that quality of output they do deserve a lot of credit cause they train the models really well and continuously manage to improve their systems to create even higher qualitive and creative outputs.

I dont think gpt4 is the biggest model that is out there but it does appear to be the best that is available.

I can run a small llm at home that is much much faster then chatgpt… that is if i want to generate some unintelligent nonsense.

Likewise there might be a way to redesign gpt-4 to run on consumer graphics card with high quality output… if you don’t mind waiting a week for a single character to be generated.

I actually think some of the open sourced local runnable llms like llama, vicuna and orca are much more impressive if you judge them on quality vs power requirement.

Kobald is a program to run local llms, some seem on par with gpt3 but normaly youre gonna need a very beefy system to slowly run them.

The benefit is rather clear, less centralized and free from strict policies but Gpt3 is also miles away from gpt3.5. Exponential growth ftw. I have yet to see something as good and fast as chatgpt

And here i though “pretty epic” seems rather weird to me. I’ll have to check on desktop later.

You should probably add a link. I already know this happend but it still reads like a joke.

https://www.dailymail.co.uk/news/article-12236093/amp/Elon-Musks-mother-Maye-says-fight-billionaire-Mark-Zuckerberg-canceled.html

Truly the weirdest timeline.

The supposed logic here is that the human genes that liked the taste of some toxic plants died out.

It can help in a survival pinch to weed out your options but I wouldn’t rely on it as a guaranteed truth.

I cannot disagree. At first i was worried about how good lemmy could possibly be with just a fraction of the reddit user base. Nowadays i feel like we got to many people already. Centralized under only a few instances no less.

At some point meta tried purchasing anonymized patient data from hospitals.

We are also discovering there are many subtle way to identify a person, like grammar, spelling speech patterns (tor specifically warns against this).

Chances are high they already have some data on you, all they need is enough subtle unique things in your data to rule out anyone else. Based on the times your most active and references you might make.

Now i am not a zuck owned superai but to give you my human attempt for an initial culling:

A social security number is the name of the American name of such number. Purposely sounding American is a strategy on itself for non Americans but the act of trying to hide your identity by pretending to be from the us can be an identifier as-well. You also seem to be more on the political left so your your likely part one of 160 mil people (for convenience i am assuming a 50% split in left/right ideology in the us)

In most countries your isp have some of your personal details your isp is also able to see all the internet domains you visit. They cant see what you do on them but because every instance has its own domain they have a timestamped list of all federated instances you visit.

It only takes that info to be leaked/hacked or sold for the big corps to obtain it. Of course if you use a vpn your isp cant see the domain but if you do have a vpn congratulations cause its yet another identifier and they too can be hacked or be selling this data.

The best way and only (still not guaranteed) way to remain anonymous is by using tor browser in windowed mode. Refraining from interacting and if you post to run your speech trough an ai algorithm to generalize/neutralize any personality from it.

As a lemmy user i can click your name and scroll trough a list of all posts and comments you have made. Thats your data and its currently free for the taking to any corporation.

Eventually instances may develop ways to protect their users from this for example make posts only visible to verified human members that are members of a community. Now imagine a user joining such community from the meta threads apps. It wouldn’t be beneath them to fulltime record the app screen and ocr all the text the user sees.

Pros: millions of users can interact with your instance

Cons: This is a super blatant attempt to consume all data. They have been doing this for years. “Log in with facebook” They see the value of the fediverse and aim to become the mainstream place for new accounts. Those users can then interact with any kind off community they want. Meta is not responsible for content on other instances. But they are harvesting all of it trough millions of clueless spies.

Time to implement a new checkbox “remember the state of the remember me checkbox”

Jokes aside this probably requires injecting some code or script into the webpage. Maybe theres a browser extension that can do this.

Lolwhat on that last line. Such a loophole.

Yeah me to. The internet is full of ads and porn as it is. What is the big deal about seeing them together?

ITT: yes, using drugs.

I cant disagree though, ive theorized that when our consciousness is in the right “state” we become more sensitive for “things we have yet to understand”

Next to make you high, drugs (and meditation, strong emotional experiences) do alter the state of our consciousness.

I am a simple man. I check if the fitgirl site i am on is the real one and thats about it.

For rare music files il try any torrent that promises to have what i am looking for. I have yet to actually catch sm malicious.

Tiling managers work great till you work with software that ignores them. (I use ms powertoys zones excessively at work for tilling and i do love em) i have yet to see one that can contain a program that is set in “fullscreen” mode. If it can do that it can probably do everything else i want.

The most straightforward way to explain it would be to have a game running in native fullscreen mode on one half of the screen but still have a fully usable and tillable desktop on the other side where you could potentially even run a second “fullscreen” program.

What i am hoping to do is to split the ultra-wide so they act as a 2 separate independent dual monitors that are connected to the same PC. As far as my own research concluded PIP/BPB are designed to use them with 2 sources, a cable each. i theorize that i can probably just plug both cables in my GPU and that will probably work but its probably not intended to be used that way and “not supported” if i run into any weird trouble.

I am pretty new to linux so i didnt know about xrandr. I’ve spend the last hour trying to get this to work just to split my 1440 monitor in 2 parts as a test. If i can get that to work than it would solve all my worries in an instant because i could just write a script to have it however i want with no extra cables attached. I haven’t gotten it to work though and even Gpt-4 admits xrandr can be rather finicky. Thanks for the tip though, i am not done trying.

Does the brain count? Id like to be give my anadamide production a boost.

First thing i try after knowing altering of brain chemicals works is probably to hunt down that mythical indigenous dmt in the pineal gland.

Use this webui (its the stabld diffusion ui for llm)

https://github.com/oobabooga/text-generation-webui

I am pretty sure it has a sever option.

Here is a list of the models it likely supports, including gpt4all. https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard

Best one i tried is wizard vicuna 13B running on a rtx2070