Are we allowed to like AI art?

[Ekco]

people already downloaded and tested it, was at like 3,000 downloads last time i checked the website, it's legit and people are already playing with the nuts and bolts of it changing the weights and measures and retraining it and other shit i don't understand

the only barrier of entry for the best LLM currently is a decent Nvidia GPU and you can run it yourself, people are coping that google and openai have better already but aren't talking about it, which doesn't really line up with their past behavior of releasing previous models without much concern for their impact on jobs or society or anything else, and it's not like the goverment is actually competent enough to regulate it even if they wanted to

ghost in the shell?

[You must be logged in to view images. Log in or Register.]
[You must be logged in to view images. Log in or Register.]
[You must be logged in to view images. Log in or Register.]

[shovelquest]

I wonder if i can use it to make a not woke LLM?

[Ekco]

Originally Posted by shovelquest [You must be logged in to view images. Log in or Register.]

I wonder if i can use it to make a not woke LLM?

you can there was a 4chan one available for awhile but they took it down off that open source model site finally iirc

guy trained it on 4chan then set it loose on 4chan

https://www.toolify.ai/ai-news/unvei...-4chan-2737836

Q: How was the AI language model trained? A: The model was fine-tuned using a dataset of 3.3 million threads from 4chan's politically incorrect board (/pol/).

3.3 million 4chan /pol/ threads is like, a non trivial amount of gamer words, prolly tens if not hundreds of megabytes of it.

[shovelquest]

Excuse my dumb questions but can you like, give it specific info to learn from, so you can raise it like a kid who never seen the outside world or something? Or is that beyond what you're able to do with it, and that's in its like lower level programming?

Like, is everything that AI learned and knows from just 4chan? or is it like getting other info elsewhere to even know how to answer or be asked a question?

[Ekco]

no you can do exactly that, and people are to a degree sorta, but like

Training: LLMs are trained on huge datasets from various sources (books, articles, code, websites).

Learning: They learn to predict the next word or code by identifying patterns.

Fine-tuning: They are further trained on specific tasks (like answering questions) to improve performance.

Alignment: Techniques like Reinforcement Learning from Human Feedback (RLHF) help align LLM behavior with human values, making their responses more helpful and harmless.
Reinforcement Learning: This is a separate process that focuses on aligning the LLM's behavior with human values and preferences. It involves:
Reward Model: Creating a model that predicts how humans would rate the LLM's responses.
Feedback Loop: The LLM generates responses, and the reward model evaluates them.

Beyond Fine-tuning: Researchers are exploring other techniques like prompt engineering to influence the behavior of LLMs without directly modifying the model itself.

most people are hacking the models by doing the Fine-tuning part from my understanding, because the "training" step 1 costs like millions of dollars in GPUs and power bill, so people aren't fuckin around as much on that step, like that step you have to explain to someone how your model is going to make money n shit lol.

but some those training data sets some are opensource and free and you can fuck around with them

[shovelquest]

thats cool thats what i want to do i hope i can learn.

this is pretty good https://x.com/FinanceLancelot/status...66609664823436

[Ekco]

yeah CNBC is pretty good at explaining this shit now because it's the "story" for wallstreet and talk about it constantly now.

was watching friday morning and it was all about ai

the insanely low cost and some of the techniques deepseek did is what the actual story is here more than it scoring a lil higher on the benchmark test, but it being opensource like everyone knows the nerdy math technique now and can copy them.
and that the american companies spent so much in comparison on their closed source private models and aren't better than the open source one is what is gonna effect the stock market

that hour long video i posted goes into it because deepseek published everything in a paper but its high level nerd shit on a whiteboard i didn't really get

[shovelquest]

This deepseek thing sounds like, it's close to iphone for devs. Like anyone can make a Ai product now...

Give me an App store, of AI software.

Integrate AI into all your PC appliactions.

Games?

[Ekco]

yeah it being able to turn someones english into code that works is the crazy insanely disruptive part

like right now its takes someone who doesn't know how to code longer than someone who does, obviously but the person who doesn't know how to code, doesn't know how to do art with enough time can bang away and create whatever the thing is they want basically,

[Ekco]

[You must be logged in to view images. Log in or Register.]
[You must be logged in to view images. Log in or Register.]
[You must be logged in to view images. Log in or Register.]

this is either gonna be a giant nothing burger buy the dip or quite a fucking around meets finding out kinda day. if it trips the circuit breaker right after open like during the start of covid the panic should set in, i hope.