News for hackers. Stuff that matters.

gabucino

gabucino on Hacker News

Re:Claude Fable 5.2 and Mythos 5.2 on Tuesday November 10, 2026 @02:01AM (#50436988)
sir, Twin Kingdom Valley does exactly this on a ZX Spectrum
Re:MaxxBench – "Nice DeepSWE score. But is it true?" on Sunday November 08, 2026 @02:25AM (#50411133)
>You can not do real code work with open source models right now

uh-huh yeah, perhaps you can't

https://github.com/whoreson/picolm

Re:Sshole on Thursday November 05, 2026 @10:14AM (#50380124)
rich felker was such a f* (funny) guy back in the mplayer days too

never changed i guess

Re:AI 'godfather' LeCun has 'zero concerns' about human extinction on Sunday October 04, 2026 @07:42AM (#49953014)
ctrl-f spacial

hn, never change

Show HN: PicoLM v1.0-rc2 ("Yura Kana"). Run an LLM on Digital Unix

Posted by gabucino on Wednesday September 23, 2026 @12:12PM
from the show-hn dept.

https://github.com/whoreson/picolm/

PicoLM is an LLM inference engine written in C99. It currently supports llama-2, GPT-2, Qwen 3.6/3.8(+MoE) and Gemma-3n models. Significant amount of work went into CPU SIMD acceleration/testing/correctness, and wide cross-platform availability with constant testing to never lose portability (from DOS through OS/X 10.4 to modernity). CUDA/HIP is supported, and accelerated IMMA kernels are...

( Read More... | no comments | on HN )

Show HN: PicoLM v1.0-rc1

Posted by gabucino on Thursday September 03, 2026 @04:19AM
from the show-hn dept.

https://github.com/whoreson/picolm/

PicoLM is an LLM inference engine written in C99. It currently supports llama-2, GPT-2, Qwen 3.6/3.8(+MoE) and Gemma-3n models. Significant amount of work went into CPU SIMD acceleration/testing/correctness, and wide cross-platform availability with constant testing to never lose portability (from DOS through OS/X 10.4 to modernity). CUDA/HIP is supported, and accelerated IMMA kernels are...

( Read More... | no comments | 3 points | on HN )