yogthos

joined 5 years ago
[–] yogthos@lemmy.ml 8 points 3 months ago (1 children)

Confederate racists who pine for the days of slavery certainly are dumb and funny.

[–] yogthos@lemmy.ml 7 points 3 months ago

A great visual illustration of the fact that police primarily exists to protect capital.

[–] yogthos@lemmy.ml 3 points 3 months ago

Yeah, pi3 isn't quite there yet to drive a laptop. I expect RISCV to mature rapidly as well. There's going to be a ton of money poured into it, and it's always easier to do things the second time around. Apple has done a lot of the hard work designing the architecture o M series chips, and I imagine a lot of it will inspire RISCV designs now. This project in particular seems pretty promising as it specifically aims to deliver high performance designs https://github.com/OpenXiangShan/XiangShan

[–] yogthos@lemmy.ml 9 points 3 months ago (3 children)

wonder if they use riscv chips

[–] yogthos@lemmy.ml 5 points 3 months ago

It's worth noting that Lemmy has plenty of different instances where opinion varies widely, including heavily liberal ones. I also don't believe in free speech absolutism because some views, like fascism, are objectively harmful. The way I look at it is the real issue with echo chambers comes from people becoming divorced from reality because they only interact with people who hold identical views. I don't see this being a major problem on Lemmy instances I frequent.

[–] yogthos@lemmy.ml 10 points 3 months ago (2 children)

And by that you mean that the views here diverge from the echo chamber you're comfortable living in.

[–] yogthos@lemmy.ml 15 points 3 months ago (4 children)

The big difference is that commercial platforms are opaquely curated by corporate interests who decide what people see, what content is allowed, and so on.

[–] yogthos@lemmy.ml 6 points 3 months ago (1 children)

Today I learned that the EU is older than the US actually!

[–] yogthos@lemmy.ml 1 points 3 months ago

Chuds living under capitalism, blaming communism for their underdevelopment will never stop being hilarious.

[–] yogthos@lemmy.ml 32 points 4 months ago (1 children)
 
 
 

Unsloth introduces reasoning capabilities in their platform using Group Relative Policy Optimization. GRPO allows users to transform standard models into reasoning models locally with as little as 7GB VRAM. Previously, GRPO was only supported for full fine-tuning, but now it works with QLoRA and LoRA.

It optimizes responses efficiently without requiring a value function, unlike Proximal Policy Optimization. Use cases for GRPO include creating customized models with rewards or generating reasoning processes for input-output data.

67
Yes (lemmy.ml)
 
111
Relatable (lemmy.ml)
 
 
view more: ‹ prev next ›