TechTakes

Running llama-2-7b-chat at 8 bit quantization, and completions are essentially at GPT-3.5 levels on a single 4090 using 15gb VRAM. I don't think most people realize just how small and efficient these models are going to become.

[cut out many, many paragraphs of LLM-generated output which prove… something?]

my chatbot is so small and efficient it only fully utilizes one $2000 graphics card per user! that’s only 450W for as long as it takes the thing to generate whatever bullshit it’s outputting, drawn by a graphics card that’s priced so high not even gamers are buying them!

you’d think my industry would have learned anything at all from being tricked into running loud, hot, incredibly power-hungry crypto mining rigs under their desks for no profit at all, but nah

not a single thought spared for how this can’t possibly be any more cost-effective for OpenAI either; just the assumption that their APIs will somehow always be cheaper than the hardware and energy required to run the model

A couple artefacts from my personal pocket of dislike for the company:

Google dot com used table layout components till feb 2022 - something that has been semantically incorrect since forever.

Google's Web.dev, a stealth advertising project disguised as a developer community, has poor accessibility test results—on AXE and it's own Lighthouse test—where developer.mozilla.org scores 100% on Lighthouse and passes with minor issues in AXE tests.

RationalWiki is a highly biased cancel community which has attacked people like Scott Aaronson and Scott Alexander before.

Background on the authors according to a far-left website.

Let's at least be honest.

That is profiling work. (Not just "Ad hominem".)

The clash with the name "rational-wiki" is too strong not to be noted.

as the infrastructure admin of a highly biased far-left cancel community that attacks people like Scott Aaronson and Scott Alexander: mmm delicious

for bonus sneers, see the entire rest of the thread for the orange site’s ideas on why they don’t need therapy:

I was about to start psychotherapy last month, I ask my family's friend therapist If he could recommend me where to go. So he interviewed me for about 30 mins and ask me about all my problems.

A week later he send me the number of the therapist. I didnt write her yet, I think I dont need it as badly as before.

Those 30 mins were key. I am highly introspective and logical, I only needed to orderly speak my problems.

to quote Key & Peele: motherfucker, that’s called a job

Saw this last night but decided to give them a few hours to backtrack. Surprisingly, they've decided to leave their comments intact!

This sort of attitude, not directly harassing trans folks but just asking questions about their moral fiber indirectly, seems to be coming from some playbook; it looks like a structured disinformation source, and I wonder what motivates them.

hey let’s see what the people who killed and buried hacker culture think should go in the jargon file!

If the spirit of the original Jargon file was to be a living document, alas, it failed to keep with the times.

Hackers at large have moved away from Lisp despite Paul Graham and other evangelists […]

Hackers also have moved away from academia at large, and 9-5 jobs at tech behemoths are more natural habitats for them, which also shaped the lingo. I mean, there’s a whole layer of slang usually pertinent to outsourcing agencies and to cubicle farms.

I can’t wait for the corporate-approved jargon file, with any hint of anti-capitalism replaced with fun words and quotes from billionaires to share as the soul leaves my body

So in order for the document to evolve, we need a system to determine consensus. Everyone who cares runs a program on their computer that joins the network and registers their intent. With each proposed change, a query goes out to the network, and it's up to everyone on the network to say yea or nay to the proposal. With enough "yea"s, the document is updated.

...this is starting to sound like a blockchain, isn't it.

for the absolute sake of fuck. coming soon: HackerDAO! collect 10xer tokens and finally prove to the junior devs why corporate gives you so many points to crunch on! vote on fun new jargon, but only if it’s crypto-related! surely you’re hacker enough to be on the pump side of this pump and dump!

OceanGate's cofounder wants to send 1,000 people to a floating colony on Venus by 2050, and says we shouldn't stop pushing the limits of innovation

https://archive.is/AyosK

this lends credence to the stealth comrade theory, btw

reposting here for better visibility (let me know if there’s a better way to do this now that we’re federated): the owners of hachyderm.io just started a generative AI for gaming project, and it looks like donations to them will likely end up going to that

Started following @techtakes and there sure are some Opinions in the replies

I talk a lot about how "empathy" in commercial UX is mostly a posture because in reality capitalism doesn't care, but it's important to consider the additional problem of people in charge who are too shallow to be capable of understanding "why" some people prefer, or need, to do things differently than they do.

This one time I was telling the ceo/founder of a startup I worked for that our react app was making my new macbook pro crawl and we need to fix that because it was a b2b product that would be used by people in finance offices decked out with dell opticrap machines. He responded with surprise "wow, steve. you really care about people don't you?"

I was kinda floored. Anyway, here we are...

https://web.archive.org/web/20230727121010/https://twitter.com/elonmusk/status/1684491212219359232

"The sad thing is that if the officer had not made a few key missteps … he might have covered his bases well enough to avoid consequences." Yeah, so sad.

For bonus sneer, check out their profile.

It's because there is a vast concerted effort by the political left to destroy Musk now that he is no longer regarded as being strictly on their side. It's at Trumpian scale these days, in terms of the venom being directed at him.

Reddit is overflowing with non-stop Musk hatred. They spout lie after lie about him in every single thread where he's a topic. The most popular lie being that his business efforts - ie his success - were funded by an emerald mine that his father owned (neither thing is true).

I say this as an emotionally independent, objective observer of the craziness, I have no stake in it, and don't feel one way or another about it. The seeming mental illness the topic of Musk seems to draw out of people is astounding however, the herd promptly acts like deranged lunatics when he comes up as a topic.

Until Trump I had never seen anything like it before, in person or online. There must be a name for such a massive scale of crowd insanity, to describe the frothing-at-the-mouth irrationality.

my deranged lunatic hivemind venomous mentally ill frothing-at-the-mouth leftist brain can’t handle how rational this fucking asshole’s take on musk and trump is

there’s also this at the top of the thread:

I've been mostly ambivalent about the Musk-era at Twitter—mostly because I just don't care enough to have an opinion.

This, though. This one makes me angry and disappointed.

Twitter has had such a solid brand for so long. It's accomplished things most marketers only dream of: getting a verb like "Tweet" into the standard lexicon is like the pinnacle of branding.

turning one of the most popular sites on the internet into a cesspit of transphobia and nazis: ambivalent

musk fails to appreciate the Twitter brand: angry and disappointed

remember, hacker news is a bastion of high-quality discussion. fucking shitheads

the API is called Web Environment Integrity, and it’s a way to kill ad blockers first and a Google ecosystem lock-in mechanism second, with no other practical use case I can find

Wolfram’s post is fucking interminable and consists of about 20% semi-interesting math and 80% goofy shit like deciding that the creepy (to Wolfram) images in the AI model’s probability space must represent how aliens perceive the world. to my memory, this is about par for the course for Wolfram

the orange site decides that the reason why the output isn’t very interesting is because the AI isn’t a robot:

What we see from AI is what you get when you remove the "muscle module", and directly apply the representations onto the paper. There's no considering of how to fill in a pixel; there's just a filling of the pixel directly from the latent space.

It's intriguing. Also makes me wonder if we need to add a module in between the representational output and the pixel output. Something that mimics how we actually use a brush.

this lack of muscle memory is, of course, why we have never done digital art once in the history of humanity. all claims to the contrary are paid conspirators in the pocket of Big Dick Blick

Of course, the AIs can't wake up if we use that analogy. They are not capable of anything more than this state right now.

But to me, lucid dreaming is already a step above the total unconsciousness of just dreaming, or just nothing at all. And wakefulness always follows shortly after I lucid dream.

only 10x lucid dreamers wake up after falling asleep

we can progressively increase the numerical values of the weights—eventually in some sense “blowing the mind” of the network (and going a bit “psychedelic” in the process)

I wonder if there's a more exact analog of the action of psychedelics on the brain that could be performed on generative models?

I always find it interesting how a hero dose of LSD gives similar visuals to what these image AI's do to achieve a coherent image.

[more nonsense]

I feel like the more we get AI to act like humans, and the more those engineers and others use LSD, the more convergence we are going to have with curiosity and breakthroughs about how we function.

the next time you’re in an altered state, I want you to close your eyes and just imagine how annoyed you’d be if one of these shitheads was there with you, trying to get you to “form a BCI” or whatever by typing free association words into ChatGPT

the only paragraph you need:

Bryan Johnson — who also has said he uses a machine to count his nighttime erections and has taken steps to make his rectum perform like a teenager’s — made his fortune in his 30s when he sold his payment processing company Braintree Payment Solutions to EBay for $800 million in cash.

this is the guy who also uses his son as his personal blood boy

NY Post, so have the archive link: https://archive.is/ExtBO

PayPal has a lot to fucking answer for

This is an infrequent sub for Silicon Valley dumbassery, Sam Altman or Paul Graham or HN reactionaries or the rest of the spawn of the Californian Ideology who aren't actually the LessWrongers.

Originally an offshoot from SneerClub, created due to off-topic orange-sitery.