Are we allowed to like AI art?

[shovelquest]

There is no such thing as lib or conserve.

neo libs, are the same thing reagen people were, is the same thing bush lovers were, is the same thing clinton fans are on and on.

Neo libs, are just "normie americans" there are Normies, and there are weridos.

Normies can't stand wierdos and call them chuds or commies (depending on what kind of t shirt the normie wears)

the sentement is the same, weather you are a chud, or a commie, (GTFO of this country if you dont like it) is the same response.

There is no left or right.

there are normies who are like bee's in a beehive.

And there are weirdos, who are hippies, and punks, and neo nazis, and the rest of the freaks and shit that keep us from all handing over our freedom to the queen.

giving everything to the queen is something BOTH the chuds and commies would do.

What political party is ted kazinski? THe progressives who deny delay depose? or the fascists that want to blame the immigrants?

It's all the same.

Normie vs wierdos.

Not left vs right. con vs lib, etc.

/essay

[Reiwa]

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

There is no such thing as lib or conserve.

neo libs, are the same thing reagen people were, is the same thing bush lovers were, is the same thing clinton fans are on and on.

Neo libs, are just "normie americans" there are Normies, and there are weridos.

Normies can't stand wierdos and call them chuds or commies (depending on what kind of t shirt the normie wears)

the sentement is the same, weather you are a chud, or a commie, (GTFO of this country if you dont like it) is the same response.

There is no left or right.

there are normies who are like bee's in a beehive.

And there are weirdos, who are hippies, and punks, and neo nazis, and the rest of the freaks and shit that keep us from all handing over our freedom to the queen.

giving everything to the queen is something BOTH the chuds and commies would do.

What political party is ted kazinski? THe progressives who deny delay depose? or the fascists that want to blame the immigrants?

It's all the same.

Normie vs wierdos.

Not left vs right. con vs lib, etc.

/essay

alot of neolibs aren't neolibs, if that makes any sense. It's just a skinsuit.

[Ekco]

progressives and trump supporters quantitatively have more in common on domestic populous polices and foreign policy than neolibs have in common with progressives.

horse shoe theory isn't just a meme

neolib / neocon is the sane middle

[shovelquest]

I always thought I was a progressive because I was like "just let them get married and stay out of their lives and stop pushing your religion on me."

Turns out the whole time I was a libertarian.

[Ekco]

lolberts have the moral correct stance with NAP and all that shit but it doesn't work in reality, their entire ideology falls apart when you scale it up past examples in a small town and all the neighbors just minding their own fucking business to talking about actual real shit like foreign policy in a zero sum game being played.

[shovelquest]

I think we can totally have all the social programs and live in a perpetual motion economy.

BUT.... you simply cannot have immigration and make any of it work.

Somewhere in between is where we should be.. we shouldn't be trying to "make one system work"

I wonder if that's just something humans will have to deal with until we're effectively "eradicated"... like if humans were in charge for another 4 thousand years, another 4 thousand after that, would we just have different romes and egypts? Probubly... Unless AI (or ourselves) eradicates us that is.

[Reiwa]

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

I think we can totally have all the social programs and live in a perpetual motion economy.

BUT.... you simply cannot have immigration and make any of it work.

Sure you can. Just don't let them have any.

[Ekco]

jack tweeted this, imma check it out, if its as claimed.

"an open-source, extensible AI agent that goes beyond code suggestions
install, execute, edit, and test with any LLM"

you're going to hear about news articles of script kiddies doing baaad shit with it.

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

this shit is so fucking dangerous to give everyone this kinda of power

[shovelquest]

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

this shit is so fucking dangerous to give everyone this kinda of power

First they just gave us tiktok for free, now this!

Radacalize, then empower the dissenters...

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

[Ekco]

Complete hardware + software setup for running Deepseek-R1 locally. The actual model, no distillations, and Q8 quantization for full quality. Total cost, $6,000. All download and part links below:

Motherboard: Gigabyte MZ73-LM0 or MZ73-LM1. We want 2 EPYC sockets to get a massive 24 channels of DDR5 RAM to max out that memory size and bandwidth.

CPU: 2x any AMD EPYC 9004 or 9005 CPU. LLM generation is bottlenecked by memory bandwidth, so you don't need a top-end one.

Get the 9115 or even the 9015 if you really want to cut costs

RAM: This is the big one. We are going to need 768GB (to fit the model) across 24 RAM channels (to get the bandwidth to run it fast enough). That means 24 x 32GB DDR5-RDIMM modules. Example kits:

Case: You can fit this in a standard tower case, but make sure it has screw mounts for a full server motherboard, which most consumer cases won't. The Enthoo Pro 2 Server will take this motherboard

PSU: The power use of this system is surprisingly low! (<400W) However, you will need lots of CPU power cables for 2 EPYC CPUs. The Corsair HX1000i has enough, but you might be able to find a cheaper option: HX1000i Fully Modular Ultra-Low Noise Platinum ATX 1000 Watt PC Power Supply

Heatsink: This is a tricky bit. AMD EPYC is socket SP5, and most heatsinks for SP5 assume you have a 2U/4U server blade, which we don't for this build. You probably have to go to Ebay/Aliexpress for this. I can vouch for this one: AMD SP5 CPU Heat Sink 4U H13SSL mz33-ar0 EPYC 9004 Fan Mute Dual Fan 9654/9554

And finally, the SSD: Any 1TB or larger SSD that can fit R1 is fine. I recommend NVMe, just because you'll have to copy 700GB into RAM when you start the model, lol. No link here, if you got this far I assume you can find one yourself!

And that's your system! Put it all together and throw Linux on it. Also, an important tip: Go into the BIOS and set the number of NUMA groups to 0. This will ensure that every layer of the model is interleaved across all RAM chips, doubling our throughput. Don't forget!

Now, software. Follow the instructions here to install llama.cpp https://github.com/ggerganov/llama.cpp

Next, the model. Time to download 700 gigabytes of weights from
@huggingface
! Grab every file in the Q8_0 folder here: https://huggingface.co/unsloth/DeepS...GGUF/tree/main

Believe it or not, you're almost done. There are more elegant ways to set it up, but for a quick demo, just do this.

llama-cli -m ./DeepSeek-R1.Q8_0-00001-of-00015.gguf --temp 0.6 -no-cnv -c 16384 -p "<｜User｜>How many Rs are there in strawberry?<｜Assistant｜>"

If all goes well, you should witness a short load period followed by the stream of consciousness as a state-of-the-art local LLM begins to ponder your question

not bad for around 6 grand , i forget where i posted it, must of been in the China thread but i said it would cost 13k the other night to run Deepseek locally. also 768GB of RAM is fucking crazy, most of the build cost is just those 24 sticks of ram

entire thread on x https://x.com/carrigmat/status/1884244369907278106

Distillation

In the context of Large Language Models (LLMs), distillation is a training technique used to create a smaller, more efficient "student" model by learning from a larger, more complex "teacher" model. The teacher model, which has already learned to perform a task well, provides training signals to the student. This allows the student to mimic the teacher's behavior, often achieving comparable performance with significantly reduced size and computational requirements. Essentially, distillation transfers knowledge from a large model to a smaller one.

Quantization

Quantization is a process that reduces the computational cost and memory footprint of an LLM by representing its weights and activations with lower-precision data types. Instead of using 32-bit floating-point numbers (FP32), which are commonly used to store these values, quantization might use 8-bit integers (INT8) or other lower-precision formats. This reduces the number of bits needed to store and process the model's parameters, leading to faster inference and reduced memory usage. However, extreme quantization can sometimes lead to a slight decrease in the model's accuracy, so careful balancing is required.

Q8

Q8 refers to 8-bit quantization, which is a common technique. However, extreme quantization can sometimes lead to a loss of accuracy. In this context, "full quality" implies that the Q8 quantization is being used in a way that preserves the model's original performance.