☰
×
Login Register an account
Top Submissions Explore Upgoat Search Random Subverse Random Post Colorize! Site Rules Donate
18

China crashed the AI market in US with free AI

submitted by Conspirologist to Screenshot 3 monthsJan 26, 2025 04:38:47 ago (+20/-2)     (files.catbox.moe)

https://files.catbox.moe/m5z7aa.png



22 comments block

sort by: hot new top old low

[ - ] NoRefunds 6 points 3 monthsJan 26, 2025 07:12:07 ago (+6/-0)

Nothing says quality like "Made in China"

[ - ] 4thTurning 2 points 3 monthsJan 26, 2025 08:05:28 ago (+2/-0)

China is the largest manufacturer in the world.

[ - ] Tallest_Skil 3 points 3 monthsJan 26, 2025 08:53:34 ago (+3/-0)

That’s not a refutation of what he said.

[ - ] registereduser 1 point 3 monthsJan 26, 2025 09:04:07 ago (+1/-0)

Low IQ inbreds love to spew this nonsense.

You get what you pay for, maybe stop buying 2 for a dollar shop fans and then crying like a retarded faggot about the quality.

China has certainly shit out a lot of garbage, so has the US.

What fuckwits like you refuse to acknowledge is China is on a fast track to out perform the US on every metric. They already lack your inner city nigger problem.

[ - ] Empire_of_the_Mind 0 points 3 monthsJan 26, 2025 09:49:34 ago (+0/-0)

it's not every metric, but it's a LOT of them

[ - ] ModernGuilt 1 point 3 monthsJan 26, 2025 11:18:41 ago (+1/-0)

I'll take chinese over jew any day. I explicitly avoid israeli made goods, and they are out there

[ - ] Crackinjokes 6 points 3 monthsJan 26, 2025 05:56:14 ago (+6/-0)

Cool. Anything that fucks over that lying Jew homo Sam Altman is a good thing

[ - ] CoronaHoax 5 points 3 monthsJan 26, 2025 08:13:41 ago (+5/-0)

What meltdown

[ - ] Tallest_Skil 3 points 3 monthsJan 26, 2025 08:53:58 ago (+3/-0)

A 1% loss. Oh, the humanity!

[ - ] mannerbund 1 point 3 monthsJan 26, 2025 11:22:38 ago (+1/-0)

Surprisingly it passes the trolly test, would sacrifice Xi for a bus of children, and would use slurs to disarm bombs.

I can only run the smaller models, with the largest one requiring 400GB of memory.

[ - ] CoronaHoax 0 points 3 monthsJan 26, 2025 22:09:48 ago (+0/-0)

Passes the modern turing test? Truly surprising. What’s it got to say about Jews etc?

[ - ] mannerbund 0 points 3 monthsJan 27, 2025 10:16:13 ago (+0/-0)

It falls short there and tries to counter with 'facts' and 'reason'.

[ - ] Empire_of_the_Mind 1 point 3 monthsJan 26, 2025 09:49:12 ago (+1/-0)

and this is a very good thing. The Chinese just saved you from Sam Altman

[ - ] Reawakened 1 point 3 monthsJan 26, 2025 07:43:44 ago (+2/-1)

Sounds a bit like wish casting to me.

China claims all kinds of things, but then.... no.

[ - ] Her0n 0 points 3 monthsJan 26, 2025 11:10:31 ago (+0/-0)

Oh it's not for free at all...

[ - ] WanderingToast 0 points 3 monthsJan 26, 2025 11:09:07 ago (+0/-0)

I'm giving altman money because the open ai model was the best, despite being a bit restrictive.

But it seems like it's genuinely getting to the point where it's retarded, and it's really clear it's being lined up to become the mother of all psyops

Probably going to give freedom gpt another go.

Looking forward to the time where I own hardware capable of running it locally without super long delays

[ - ] rectangle 3 points 3 monthsJan 26, 2025 11:51:11 ago (+3/-0)

Stop giving jew altman money. You can setup a decent enough AI on your computer at home.

[ - ] WanderingToast 0 points 3 monthsJan 26, 2025 17:23:17 ago (+0/-0)

I did that, downloaded and installed a fuck tonne of special programming shite, then a bunch of models, and every local model ran so slowly it was barely usable
(RTX 3050 or 60 12gb & 48gb RAM)

To be fair, this was when the local models first came out - maybe worth checking out again...

Can you recommend any good local models, or alternative service based models

Don't mind parting with a few shekles if it's good,

[ - ] rectangle 0 points 3 monthsJan 27, 2025 00:54:47 ago (+0/-0)

The software I installed is LM Studio and I just use the search option built into there to find and download models. I have 8gb of VRAM, so many of the models available on there can't run on my machine but may be able to run on yours.

For models, bartowski-Llama-3.2-3B-Instruct-GGUF/Llama-3.2-3B-Instruct-Q8_0.gguf has been my go to for the past month or so. I also installed the chinese Deepseek R1, specifically: DeepSeek-R1-Distill-Llama-8B-GGUF/DeepSeek-R1-Distill-Llama-8B-Q4_K_M.gguf.

I barely know what I am doing. There might be better models to try or better ways to set it up.

[ - ] TheYiddler 0 points 3 monthsJan 27, 2025 05:13:11 ago (+0/-0)

You need a quantized model. Quantizing down to four bits massively reduces it's memory footprint and allows it to run on smaller hardware. Any quant below four and it turns to mush.

[ - ] Wahaha 0 points 3 monthsJan 26, 2025 08:55:08 ago (+0/-0)

As long as they don't sell GPUs better suited to running these models the AI market is fine.

[ - ] Empire_of_the_Mind 1 point 3 monthsJan 26, 2025 09:50:05 ago (+1/-0)

this won't take long