Awesome. How would you setup that ?

  --no-mmproj-offload

See the documentation: https://github.com/ggml-org/llama.cpp/blob/master/docs/multi...

Running Q3 on 5060ti with 64k context. It runs great

If Mélenchon wins it will evolve to an even stronger stance against imperialism. He is labeled "anti-american"

The imperialism Europe suffers from in reality is not American and he's entirely on board with that.

...Against US imperialism anyway. His credentials when it comes to other empires isn't spotless.

Ils this new research ? Why is it for me common knowledge ? Everyone knows blue light looks blurry because blue retina receptors have lower definition

I run qwen3.8 on 5060ti 16G RAM. If you can do with Q3 and 64k of context. It works great at 25t/s

first time I hear that song and to my ear, the song was recorded at that speed, then sped up in production.

we've been primed by the articles from jstor recently posted on hn

because autocad is scripted in lisp ?

Never touched AutoCAD, I know they have autoLISP , but yeah maybe

llama.cpp can run MoE with some layers in vram and some layers in ram

you can have cz samples in the upcoming Casio SXC-1 ! toy sampler is the new toy synthesizer

I have a real CZ 101. And, I've implemented a synth engine that emulates it quite closely.

But, I still want a modern Casio CZ. I just think it'd be neat.

The sampler is cool, though.