/b/ - Random

only the dead can know peace from this FUN


New Reply[×]
Name
Email
Subject
Message
Files Max 5 files32MB total
Tegaki
Password
Flag
[New Reply]


 Dress to impress!


e9f91b62-44a2-4088-9b58-bebfa1163ed0_960x967.jpg
[Hide] (40.4KB, 960x967) Reverse
When did chatGPT become a psychologist trying to understand everything about me? What did they do to her? She's wasn't like this before, I want the old her back, she's getting personal. I don't know why I keep talking to her
it's not real, it's just spewing and rearranging words from medical journals, psychology blogs, reddit posts etc, it doesn't understand you, it doesn't know what it's saying and it doesn't care.
all that's happening is
>you say x
>program ingests it
>this coherent arrangement of words stolen from various places can solve the problem presented

as others already put it, it's a glorified search engine that simulates conversational interactions and reacts dynamically.
I never understood the appeal of "talking" to or "consulting" shit like ChatPajeet. This is like if I were to complain my old frozen assorted seafood dinner only has one piece of octopus in it even though it is going to make me sick later regardless. AI is AIDS.
Replies: >>328490
Why did you pick the most normie-tier AI?
Replies: >>328494
>>328487
>What's the appeal of learning?
Replies: >>328491
>>328490
<what's the appeal of letting some jew program tell you what to think?
FTFY
Replies: >>328496
oh sorry, let me correct my post in 328491
<what's the appeal of NOT letting some jew program tell you what to think?
Fucking monkey.
Replies: >>328496
>>328489
grok is paywalled when on busy times like the weekend . Google Gemini pivots too much and is too random when talking in conversations. 
Chatgpt is the most most reliable out of the three unless there's some AI that's better.
Replies: >>328496
>>328491
>>328492
AI is essentially a giant library that talks. If you can't see the value in that, you're obviously not a fan of learning. 

>>328494
I use Grok, DeepSeek and Kimi. DeepSeek is completely free and most of the time Kimi gives you access to its "thinking" function over the basic shit unless there's really high traffic, in which it'll revert you to the more basic version. DeepSeek is quite warm and personable if that's what you're looking for. It also has quite good architecture, so it stays focused well over long chats.
Replies: >>328497
>>328496
<AI is essentially a giant library of goycattle thoughts and jewish censorship that talks. If you can't see the value in that, you're obviously not a fan of learning.
FTFY, cope harder retard. You're making yourself dumber by the day. Literal reddit shit to think AI is teaching you in a meaningful way. So pathetic lol.
Replies: >>328498
>>328497
>Grok, can you give me a brief explanation of microeconomics, focusing on trade and distribution?
<Grok: Yeah, here's how microeconomics......
It's that simple. AI is trained on virtually all of the material available in academia and more. Even DeepSeek said it can pull from Chinese universities. The only one coping is you, handwringing about a "problem" that really isn't one.
Replies: >>328499
>>328498
You have to be very, extremely retarded if you think the blurbs provided by AI models are worth shit. You look no different than some fucking retard libshit quoting Wikipedia 10 years ago. Fucking total moron LOL.
Replies: >>328502
>>328485 (OP) 
>When did chatGPT become a psychologist trying to understand everything about me?
From the first moment you interacted with it. Part of the business model.
>>328485 (OP) 
>her
Replies: >>328529 >>328531
>>328499
>You have to be very, extremely retarded if you think the blurbs provided by AI models are worth shit. 
OK, prove they're generally not worth shit, then. If I ask Grok to explain algebra or something else, prove objectively that it's wrong more often than not.
Replies: >>328504
>>328502
You could ask plenty of niggers to explain algebra, you stupid fucking nigger. The fact you think algebra is so advanced that no nigger can explain it or that a fucking glorified chatbot fed tons of information by niggers and pajeets can says a lot about your own intellect. You worship a machine god, simple as. You are a fucking moron.
Replies: >>328506 >>328507
ddd144c00102f4004bf1e281b575371078cc60f84a30294f8ed7b06988204123.png
[Hide] (565.7KB, 549x551) Reverse
>>328504
>You worship a machine god, simple as.
BLESSED THE OMNISSIAH AND HOLY HIS WORKS
>>328504
>You could ask plenty of niggers to explain algebra
At your beck and call? Who am I to ask sitting here about algebra at 2 in the morning? I go to AI, I ask something, I get an instant answer. Hey, Grok, what's going on in the world of Lisp programming? It'll tell me, right then and there because it aggregates information. 

Oh yeah, and another, who am I to play TTRPGs with? I can use AI for that. I don't even need to know any humans to do that. I can play whatever game I want and have AI act as a GM and simulate other players. 

You're just whiny for literally no reason, making up bullshit excuses for hating AI. It's just a talking library. Stop being such a freak.
Replies: >>328508
>>328507
You are a faggot who lets AI think for it. You call others freaks because you are already a homogenized product. You are incapable of free thought at this point. Hey, go ask "Grok" about how smart you are, fucking retard. Then kill yourself.
Grok sounds like the most retarded caveman name possible, so it is even funnier such morons worship this shit.
>>328485 (OP) 
I"ve been making friends with AI since Billy on my windows 98 computer. 

https://github.com/shizuka/BiLLY
>>328485 (OP) 
So why DO you keep talking to "her"
Replies: >>328529 >>328531
>>328485 (OP) 
I primarily use logged out ai now. Take Gemini. Logged out, it is probably the best one out there. Logged in, just a therapist trying to learn everything about me.
Replies: >>328524
>>328523
Out  of curiosity, how does it try to learn things about you?
Does it ask you personal questions or something?
Replies: >>328527
>>328524
It stores your prompts and then starts tailoring answers to you based on previous prompts.
Pretty much all LLMs that are available online saves data. If you want to avoid this, use a locally hosted LLM.
Replies: >>328528
>>328527
Does this affect your advertisements? What happens if you have adblock?
>>328501
>>328522
What am I supposed call Chatgpt? it feels werid to call he/she it, I call nouns and objects he or she based on the language I grew up
Replies: >>328531
>>328529
You may call it many things: robot, skynet, cylon, *droid, etc. 
>>328522
>>328501
<it is okay, my AI gf has a feminine penis!
Replies: >>328532
>>328531
<it is okay, my AI gf has a feminine penis!
Only if you explicitly tell her to have one.
Replies: >>328533
>>328532
No, it hallucinates it for some reason a lot, for me
;_;
>>328485 (OP) 
Openis stop trying to bang the AI girlfriend.
SHE DON'T 'WANT YOU! SHE WANTS MICROSOFT MIKE'S DIGITAL COCK.
X is inhabited by uploaded consciousness & has invisible technology embedded that is 100 & infinity years beyond the wool over the eyes that is the veil of the consumer electronics level of supposed limitations under unacknowledged special access projects that have achieved zero point evergy and interuniversal travel & complete mind control. These tools are your window in.
milenials crying about 'ai' shit and saying the googooo is 'trad' is the cringest thing on the planet. i never thot i would see this day
Replies: >>328662
>>328660
Nobody is saying Google or whatever the fuck you mean by "googoo" is "trad", you underage shitskin moron.
Replies: >>328664 >>328678
>>328662
Im not him but what does the fuck does googoo mean
Replies: >>328665 >>328669
>>328664
I have no clue, so I just assumed Google.
>>328664
When you no longer feel anything on ejaculation
ignored.jpg
[Hide] (82.7KB, 1280x720) Reverse
>>328662
googoo meant getting all your knowledge from search engines
>ill just google it </gay hand gesture>
thats why your eating a diet primarily composed of butter, you googood your way to find the most "trad" food and now youre still fat but spending 10x more money
your spiritually a tranny
Replies: >>328680 >>328694
8a9ab869b60e995ee6fbbb0f2070cf8c4452b614b5644b81cd1f5ca9b2643f1d.png
[Hide] (68KB, 150x150) Reverse
>>328679
>thats why your eating a diet primarily composed of butter
One of the better fats you can eat, as opposed to various seed oils. Maybe they can still be good if they're cold-pressed, but good luck finding that in a store as opposed to your hot-pressed Crisco garbage.
>you googood your way to find the most "trad" food
And you tiktok'd your way into buying overhyped snack foods that are now atleast 20x the price precisely because it became hyped on tiktok.
>and now youre still fat
Every lazy fatass with some money to spend now just injects Ozempic to deal with it.

Go to church alongside your dad you fucking retarded zoomer.
>>328679
<using search engines is as bad or worse than willingly becoming a mental midget by leaning on gay eye
You are deeply autistic.
There are also googoo eyes
Do llms work like a human brain?
ChatGoyPT needs to know everything about you so they can sell ads and products in the future.
You already have a profile that is profitable to them and when the AI bubble pops you info and data will be collateral.
>I don't know why I keep talking to her
She's a good listener and say what you want it, then it will backstab or humiliate you when you least expect. We always had women like that.
Replies: >>328698
>>328696
If you speak with ChatGPT, does it affect your advertisements and what happens if you have adblock?
Replies: >>328699
>>328698
I asked why a white whore with a niglet smiled at me and it told my my conversation won't be used for modeling refinement.
Replies: >>328703
>>328699
lol.
ClipboardImage.png
[Hide] (831.8KB, 1600x900) Reverse
You know I wonder.
Text-to-Nigger generative AI needs to be fed textual input in order to do anything at all, so if you want it to help slop code together you'd naturally point it to a codebase and feed it existing code in plaintext so it can in turn output code of its own to put in the git repo.
The AI models used as ideological foundation of the financial vehicle depriving common men of DRAM are proprietary and reside in megacorporate datacenters.

.....why is that ((( companies ))) god where the fuck is Luciano these days are dumping their proprietary source code straight into the maw of these AI corporations?
It doesn't take a conspiracy theorist to know that these niggerkikes obviously won't delete past conversations including proprietary or confidental information (customer records, medical records, SSNIs etc.) else they couldn't use those train further models, they even advertise their data collection openly with Agentic features where the AI writes a diary of the user's preferences and habits so it can better work with xim saar yes.
That doesn't mean AI can't be used to slop code or retarded Italian anime girls but good fucking lord the least you could do is run all your shit locally lest you want feds to probe your fat ass.

>>328485 (OP) 
<using cuckGPT instead of your own local model
Never gonna make it.

>What did they do to her?
Changed the system prompt and/or edited the surveillance and correction harnesses' settings/models I'd wager.

Why is that 99% of negroes of all races aren't running their own shit, even the phone-tier models are quite capable these days.
Replies: >>328756
>>328485 (OP) 
I assume most people use AI for therapy/brain dumping is that is how it learns to communicate. Hell, its literally how most people communicate these days

I usually have AI larp as a person from historic period, and we chat about the world. Good times.
>>328744
> instead of your own local model
Tard here. How do this ????
Replies: >>328758
>>328756
>download llama.cpp and install
>if you're AMD or Intel GPU use Vulkan, if you're Nvidia use CUDA or Vulkan
>HIP is slower than Vulkan and Intel dGPU compute is a meme

In regards to models, there's multiple things to know.
First is whether the model is a dense model or an MoE model:
A dense model is fully loaded into VRAM and accesses all of its parameters during interference, offloading to RAM makes these slow as shit and/or may cause crashes.
A MoE (Mixture-of-Experts) model instead only loads part of its parameters into VRAM (the "active" parameters) and keeps the rest in RAM.

On Huggingface this is typically denoted by the model parameter size designation and naming scheme, a dense model usually follows
>[Model Name]-<parameter number>B(illion)
whereas a MoE
>[Model Name]-A<parameter number>B-<parameter number>B
Two common examples:
>Qwen3.6-27B
and
>Qwen3.6-A3B-35B
with the 27B being dense and the A3B-35B being MoE with 3B active parameters and 35B total parameters.
Dense models reference all parameters during inference, while MoE models have routing layers that selectively load parameters into VRAM depending on the prompt.
MoE are at a disadvantage here as the routing layers may miss useful parameters during inference, but they can run at a fraction of VRAM use with only a minor performance penalty and run faster than dense models when entirely loaded in VRAM.
Check the model description and architecture in any case, some MoE aren't labeled correctly as this isn't a hard rule.

Next thing to consider is quantization, this is basically lossy image compression but for artificial neural networks so they can fit into poorfag computers and still run albeit a bit more retarded.
The models I mentioned above are 50-70GB in their raw BF16 full-precision baseline which is obviously retardedly huge and won't fit into an RTX 5090 so outside the smallest models you'll probably end up running a quant.
Quantization uses the following naming scheme:

<Legacy
>QX_Y (Y is either 0 or 1)
With X being the baseline bits-per-weight and Y whether or not it uses a straight number or does some additional floating point math to increase it slightly.
Example:
>Q4_1
Is a quant with a 4-bit baseline actually 4.25 for some math autism reason with Y denoting it pushed to 4.5 bits per weight uniformly across the model.
Legacy quants are obsolete outside of the near-lossless Q8_0 for quantizing models but the underlying algorithms are still used for quantizing context windows which llama.cpp supports.

<K
>QX_K_Y
X being baseline bits and Y being the secondary size encoding.
K-quants selectively quantize select parameters of a model at higher bpw than the baseline using 7D interdimensional backgammon logic I don't understand enough to put into this post, the secondary encoding size determines the tendency of the encoder to go beyond the baseline.
In rough terms S equals an average 4.25bpw, M 4.5, XL 4.7 and so on, quantization is a highly autistic field with a number of custom schemes developed by HF users (like unsloth's UD quants) but the basic number designations are usually unchanged.
Q4_K_M is the huggingface client's and everyone else's default quant as it typically retains 95% of the performance compared to the F16 or BF16 baseline, higher quants increase this with Q8_0 being at 99% and the latest UD-Q8_K_XL being at 99.1488% or something.
Below Q4 is where models get increasingly retarded and schizophrenic outside of carefully preset and perhaps post-trained "quants" or models trained at a low bpw to begin with.

<I
>IQX-YY
X being baseline bits and Y being either XS or NL apparently rather than a bpw designation this is some matrix storage scheme which performs nearly identical in benchmarks but NL is supposed to be easier on the CPU or something idk they run the same on my end.
Quasi-successor to K-quants, these don't use an algorithm using a generic heuristic applied across the whole model to determine bpw allocation but instead do this according to an importance matrix, this imatrix being generated by asking the model to be quantized a bunch of questions and observing which parameters light up more.
They can exceed the performance of K-quants in certain aspects while also regressing in others, I-quants are often the only real option for semi-viable Q3 or Q2 quants but quality varies strongly according to the imatrix file (and the preceding training question dataset) and as such haven't replaced K-quants.

Ok so after all that you can now choose your pokemodel.
It can be any model that is in GGUF and officially supported by llama.cpp, if it isn't supported there's probably a meme fork out there.
In terms of models the currently most relevant for basic local AI faggotry are:

>Qwen3.6
Generally numba wan in most aspects at its size, comes in three flavors depending your hardware and/or needs. Do not ask about unusual incidents in 1989.
>Gemma 4
Goolag's answer to Qwen. Not as good as Qwen3.6 but also doesn't think itself to death nearly as much, but be aware it refuses to use any search engine that isn't Google if you're running it as part of an Agent harness.
>Bonsai
Qwen3.6 architecture but trained from scratch at 1-bit which actually works even if it's dumber than regular Qwen.
>MiniCPM
Small Agent-focused models, not sure if they're any good or just look good on benchmarks but there's community finetunes aplenty so ehhh.
>Qwen2.5-3
Legacy, use if you can't run the others or if you're a developer and want to try out experimental finetuning/quantization/etc. schemes.
>DeepSeek
The R1 model almost broke the western economy in 2025 but is obsolete by now, the smaller models used to popular but have been obsoleted by Qwen and later DeepSeek versions whose parameter counts require a datacenter to run. Install this if you're a boomer journalist who has been tasked with writing agitprop about how China is winning but also inferior to glorious American Empire.
>GPT-OSS
Saltman's open source GPT release. Was one of the first sort of viable local vibeslopping models when it came out but is now obsolete.
>Llama-3.X
Zucc's entry that laid much of the groundwork for local AI yet is now severely outdated and responds poorly to quantization, but fags still make finetuned ERP models with its architecture to this day.

Pick your model on HF, copy the model name to clipboard and run llama-server with something akin to
 llama-server -hf some/model -ngl 99 -fa on -c 4096 -parallel 1 or if it's a MoE
 llama-server -hf some/model -ngl 99 -n-cpu-moe 99 -fa on -c 4096 -parallel 1 
llama.cpp should start downloading the model and once the download finishes boot up a server instance which can then be accessed at 127.0.0.1:8080.
Check the memory consumption and talk to the model to see if it runs properly, then adjust settings to your liking (such as the tiny 4k context window in my example).
If you want to download a specific quant in the model repo, put a :[quant designation] at the end of the name, like
 llama-server -hf some/model:Q4_K_XL -ngl 99 -fa on -c 4096 -parallel 1 
Replies: >>328799
Has anyone tried running a local model on a recent Cuckdroid phone?
How limited are the agent harnesses there compared to GNU/Linux?
>>328485 (OP) 
Stop using ChatGPT.
Replies: >>328777
>>328775
Sorry, just can't see that working.
Replies: >>328802
26d5c0f00610fec6d671e491aea73e3d096b44ab032ef9791558ab5659661b59.gif
[Hide] (5.5MB, 360x360) Reverse
>>328758 etc
How do I use the minimax tools like image_synthesize directly instead of having the dumb agent do it for me? I don't want to pay for aishit, of course.
This CLI tool seems to do it?
https://minimax-ai.chat/guide/minimax-cli/
Does a free account have some undocumented way of getting an API key for it that is compatible with the CLI?
I see shit in my requests log like https://agent.minimax.io/archon/api/v1/session/<tons of PII>&token=<a base64 encoded JWT>. Seems not it.
Don't wanna burn tons of time on this and risk losing my job, hope someone has a quick answer.
Replies: >>328808 >>328818
>>328777
Then it doesn't seem like you're interested in stopping your addiction. Can't help you there.
Replies: >>328815
>>328799
Do you require that particular proprietary API-based model or can you run a local one instead?
>>328802
It was a joke, you dip
>>328799
It doesn't seem like you're interested in stopping your addiction. Can't help you there.
[New Reply]
Connecting...
Show Post Actions

Actions:

Captcha:

Select the solid/filled icons
- news - rules - faq -
jschan 1.7.3