The moral of this story?

Biomagnification.

https://en.wikipedia.org/wiki/Biomagnification

The sword is backup. This guy knows science can go wrong.

Can confirm. Divers freak out, though I don’t understand why.

But the orca is fake. 100% never been eaten by one.

My level of worry hasn’t lowered in years…

But honestly? Low on the totem pole. Even with Trumpy governments.

Things like engagement optimized social media warping people’s minds for profit, the internet outside of apps dying before our eyes, Sam Altman/OpenAI trying to squelch open source generative models so we’re dependent on their Earth burning plans, blatant, open collusion with the govt, everything turning into echo chambers… There are just too many disasters for me to even worry about the government spying on me.

If I lived in China or Russia, the story would be different. I know, I know. But even now, I’m confident I can given the U.S. president the middle finger in my country, but I’d really be more scared for my life in more authoritarian strongman regions.

Especially since it means there could be more.

Ants too.

I’ve learned to love bees. They can ride in my car. Maybe even some spiders.

Hornets, though? Big fat nope.

Presumably you will advance along with humanity though, or failing that, just figure out the transcendence thing yourself with so much time?

I don’t think anyone would choose to stay ‘meatbag human’ for trillions of years.

Almost all of Qwen 2.5 is Apache 2.0, SOTA for the size, and frankly obsoletes many bigger API models.

To actually answer this, you could look into free APIs of open source models, which have daily limits but are otherwise largely catch-free. You could even mirror endpoints on your VPS if you need to, or host “middleware” like prompt formatters and enhancers.

I say this because, as others said, you cannot actually host AI on a VPS…

I think there’s also a “Netflix effect” where old games are incresingly accessible as an alternative to newer crap, kinda like (from my personal observations) how a lot of young people seem to be really fluent in old movies and TV due to streaming and YT.

Its going to bite these publishers in the bum.

It’s unbelievavly time inefficient for… anything.

And its incredibly engaging. I burnt through so much time shooting the breeze in hopes of actually finding something interesting, notification spam, checking channels… It’s why I deleted it from everywhere. And it left a gaping hole in my life, because its the only place some niche communities exist now.

It turns out the popular alternative is “force you to sign up (with a phone number) from critical mass/FOMO, track the snot out of you then slide ads in later.” Oh, and the stuff you want is siloed away until you join, and buried in a mountain of rambling and engagement optimization junk.

Note that I’m largely talking about Discord, which is unfortunately where many of my interests have been shunted off to. People talk about Facebook, Google and OpenAI eating the internet, but I feel like Discord is the quiet trojan horse.

These days, there are amazing “middle sized” models like Qwen 14B, InternLM 20B and Mistral/Codestral 22B that are such a massive step over 7B-9B ones you can kinda run on CPU. And there are even 7Bs that support a really long context now.

IMO its worth reaching for >6GB of VRAM if LLM running is a consideration at all.

I am not a fan of CPU offloading because I like long context, 32K+. And that absolutely chugs if you even offload a layer or two.

For local LLM hosting, basically you want exllama, llama.cpp (and derivatives) and vllm, and rocm support for all of them is just fine. It’s absolutely worth having a 24GB AMD card over a 16GB Nvidia one, if that’s the choice.

The big sticking point I’m not sure about is flash attention for exllama/vllm, but I believe the triton branch of flash attention works fine with AMD GPUs now.

Basically the only thing that matters for LLM hosting is VRAM capacity. Hence AMD GPUs can be OK for LLM running, especially if a used 3090/P40 isn’t an option for you. It works fine, and the 7900/6700 are like the only sanely priced 24GB/16GB cards out there.

I have a 3090, and it’s still a giant pain with wayland, so much that I use my AMD IGP for display output and Nvidia still somehow breaks things. Hence I just do all my gaming in Windows TBH.

CPU doesn’t matter for llm running, cheap out with a 12600K, 5600, 5700x3d or whatever. And the single-ccd x3d chips are still king for gaming AFAIK.

To go into more detail:

• Exllama is faster than llama.cpp with all other things being equal.

• exllama’s quantized KV cache implementation is also far superior, and nearly lossless at Q4 while llama.cpp is nearly unusable at Q4 (and needs to be turned up to Q5_1/Q4_0 or Q8_0/Q4_1 for good quality)

• With ollama specifically, you get locked out of a lot of knobs like this enhanced llama.cpp KV cache quantization, more advanced quantization (like iMatrix IQ quantizations or the ARM/AVX optimized Q4_0_4_4/Q4_0_8_8 quantizations), advanced sampling like DRY, batched inference and such.

It’s not evidence or options… it’s missing features, thats my big issue with ollama. I simply get far worse, and far slower, LLM responses out of ollama than tabbyAPI/EXUI on the same hardware, and there’s no way around it.

Also, I’ve been frustrated with implementation bugs in llama.cpp specifically, like how llama 3.1 (for instance) was bugged past 8K at launch because it doesn’t properly support its rope scaling. Ollama inherits all these quirks.

I don’t want to go into the issues I have with the ollama devs behavior though, as that’s way more subjective.

It’s less optimal.

On a 3090, I simply can’t run Command-R or Qwen 2.5 34B well at 64K-80K context with ollama. Its slow even at lower context, the lack of DRY sampling and some other things majorly hit quality.

Ollama is meant to be turnkey, and thats fine, but LLMs are extremely resource intense. Sometimes the manual setup/configuration is worth it to squeeze out every ounce of extra performance and quantization quality.

Even on CPU-only setups, you are missing out on (for instance) the CPU-optimized quantizations llama.cpp offers now, or the more advanced sampling kobold.cpp offers, or more fine grained tuning of flash attention configs, or batched inference, just to start.

And as I hinted at, I don’t like some other aspects of ollama, like how they “leech” off llama.cpp and kinda hide the association without contributing upstream, some hype and controversies in the past, and hints that they may be cooking up something commercial.

Nah, I should have mentioned it but exui is it’s own “server” like TabbyAPI.

Just run exui on the host that would normally serve tabby, and access the web ui through a browser.

If you need an API server, TabbyAPI fills that role.

Shrug did you grab an older Qwen GGUF? The series goes pretty far back, and its possible you grabbed one that doesn’t support GQA or something like that.

Doesn’t really matter though, as long as it works!

Your post is suggesting that the same models with the same parameters generate different result when run on different backends

Yes… sort of. Different backends support different quantization schemes, for both the weights and the KV cache (the context). There are all sorts of tradeoffs.

There are even more exotic weight quantization schemes (ALQM, VPTQ) that are much more VRAM efficient than llama.cpp or exllama, but I skipped mentioning them (unless somedone asked) because they’re so clunky to setup.

Different backends also support different samplers. exllama and kobold.cpp tend to be at the cutting edge of this, with things like DRY for better long-form generation or grammar.