Yes to soyjaks, hard no to ai
GenZedong
This is a Dengist community in favor of Bashar al-Assad with no information that can lead to the arrest of Hillary Clinton, our fellow liberal and queen. This community is not ironic. We are Marxists-Leninists.
This community is for posts about Marxism and geopolitics (including shitposts to some extent). Serious posts can be posted here or in /c/GenZhou. Reactionary or ultra-leftist cringe posts belong in /c/shitreactionariessay or /c/shitultrassay respectively.
We have a Matrix homeserver and a Matrix space. See this thread for more information. If you believe the server may be down, check the status on status.elara.ws.
Rules:
- No bigotry, anti-communism, pro-imperialism or ultra-leftism (anti-AES)
- We support indigenous liberation as the primary contradiction in settler colonies like the US, Canada, Australia, New Zealand and Israel
- If you post an archived link (excluding archive.org), include the URL of the original article as well
- Unless it's an obvious shitpost, include relevant sources
- For articles behind paywalls, try to include the text in the post
- Mark all posts containing NSFW images as NSFW (including things like Nazi imagery)
hard no to ai
Why? Don't want to start another runaway debate thread about AI, but I assume you have a reason and want to know what it is before I jump to conclusions.
High resource consumption for low return, burning all that electricity to produce something you could draw in MS paint in 20 minutes is completely unjustifiable
You use more electricity streaming an episode of a HD netflix show than generating like 100 text prompts. Image generation is also way less costly than streaming.
The only really costly thing with LLMs is the training stage.
I was wondering if it was an environmentalist thing. That's understandable. I don't personally think it's unjustifiable in the larger context of things and the fact that those data centers are being pushed hard whether individuals want it or not, as capitalists rush to get in on the ground floor of investment hype; so some of it arguably comes down to that thing of, using the tools that are already there anyway. And over time, compute is getting less intensive for more; I'm also pretty sure compute for image tends to be significantly less than that of text (counter-intuitive as that may sound, I think it's to do with the size of models).
But I am not trying to persuade you to use AI. Just explaining why I don't take the same position.
Fair
I mean training a model is incredibly expensive both in electricity and labor, but none of us are doing that. Is using already-trained models locally that expensive?
I don't have a GPU so I haven't delved into this, @yogthos@lemmygrad.ml what's your take?
You're correct, once a model is trained, actually running it isn't that expensive.
Hell yeah. I love 'jaks. Hear something you don't like? Quote 'em with a regular soyjak. «Huh? That's all you can do? Quote me with a soyjak? Fucking zoomers.» Quote them with an angry, crying soyjak.