Lightnews — Scholar-powered news

The Hybrid Group

@hybridgroup.com

We're moving at the speed of thought, so yzma v1.0 beta2 is out!

Better, faster, and more benchmarks to show it too.

Run local models using Go with your CPU, CUDA, or Vulkan.

You know what to do!

github.com/hybridgroup/...

#golang #llama #llamacpp

GitHub - hybridgroup/yzma: Go for hardware accelerated local inference with llama.cpp directly integrated into your applications

Go for hardware accelerated local inference with llama.cpp directly integrated into your applications - hybridgroup/yzma

github.com

November 24, 2025 at 2:51 PM

The Hybrid Group

@hybridgroup.com

yzma 1.0 beta1 is out!

Use Go for hardware accelerated local inference with llama.cpp directly integrated into your applications. No external model servers or CGo.

Go get it right now!

github.com/hybridgroup/...

#golang #llama #vlm #llm #local #gpu

GitHub - hybridgroup/yzma: Go for hardware accelerated local inference with llama.cpp directly integrated into your applications

Go for hardware accelerated local inference with llama.cpp directly integrated into your applications - hybridgroup/yzma

github.com

November 20, 2025 at 9:14 PM

Reposted by The Hybrid Group

William (Bill) Kennedy

@goinggo.net

Thanks to @deadprogram.com and his Yzma project, you don't need to deploy model servers anymore, you can run GGUF models directly in your #golang code.

I have cool examples, including a full RAG app using DuckDB. I will have more complex examples soon.

github.com/ardanlabs/ai...

ai-training/cmd/examples/example13 at main · ardanlabs/ai-training

Provide examples for Go developers to use AI in their products - ardanlabs/ai-training

github.com

November 16, 2025 at 3:36 PM

Reposted by The Hybrid Group

takasago

@sago35tk.bsky.social

#BuriKaigi 2026 内で開催する TinyGo Keeb Tour (はんだ付け + ソフトなワークショップ) で使う zero-kb02 を準備していってる。みんな、はんだ付けしにきてね！今は TinyGo のファームウェアしかないけど、 Vial とか zmk とか prk でファームウェア書いてくれる人も募集中です。
#tinygo_keeb

November 14, 2025 at 12:21 AM

Reposted by The Hybrid Group

deadprogram

@deadprogram.com

"Captions With Attitude" in your browser from your webcam generated by a Vision Language Model (VLM) from a Go program running entirely on your local machine using llama.cpp!

github.com/hybridgroup/...

#golang #vlm #openCV #llama #yzma

November 11, 2025 at 8:24 PM

The Hybrid Group

@hybridgroup.com

Life's comes at you fast, and so do new releases of yzma!

Use pure Go for hardware accelerated local inference on Vision Language Models & Tiny Language Models.

0.9.0 out now with API improvements, model downloading, & more.

github.com/hybridgroup/...

#golang #llama #vlm #tlm

GitHub - hybridgroup/yzma: yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo.

yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo. - hybridgroup/yzma

github.com

November 7, 2025 at 4:59 PM

Reposted by The Hybrid Group

Ardan Labs

@ardanlabs.com

Ult. Software Design LIVE Schedule

Join @goinggo.net & @kenriquezcodes.bsky.social in this week's streams:
Tue. 11/4 & Thu.11/6 from 11am - 1pm EST

Tomorrow we'll have a special guest: @deadprogram.com

Stay tuned 1hr before the LIVE show for the stream link!😎
📽️Last episodes here: bit.ly/3CShDOS

November 3, 2025 at 8:31 PM

The Hybrid Group

@hybridgroup.com

yzma 0.8.0 is out, now with over 87% coverage of the llama.cpp API from pure Go! More robust, more examples.

Go get it right now!

github.com/hybridgroup/...

#golang #llamacpp #vlm #slm #tlm

GitHub - hybridgroup/yzma: yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo.

yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo. - hybridgroup/yzma

github.com

November 3, 2025 at 10:38 AM

Reposted by The Hybrid Group

TinyGo

@tinygo.org

On October 12th, the first-ever TinyGo Conf happened in Tokyo, Japan. Report from team member Daniel Esteban aka "Conejo" tells all!

#tinygo #tinygoconf #golang #japan

madriguera.me/tinygo-conf-...

TinyGo Conf 2025 JAPAN

On October 12th, the first-ever TinyGo Conf happened in Tokyo, Japan. I planned to write about it sooner, but I did so many things during my trip that I didn't have the time nor strength to do it unti...

madriguera.me

October 28, 2025 at 6:04 PM

Reposted by The Hybrid Group

David Aronchick

@ironyuppie.com

In which, sadly, I continue to strongly agree with @squillace.bsky.social www.distributedthoughts.org/webassembly-...

WebAssembly 3.0 and the Infrastructure We Actually Need

"We've finished the model... when do we get to the value?" Here's what I'm watching happen in real time: DevOps teams are spending $$$$$$ annually on cloud egress fees just to move ML models between ...

www.distributedthoughts.org

October 23, 2025 at 6:11 PM

Reposted by The Hybrid Group

deadprogram

@deadprogram.com

yzma has a logo now!

github.com/hybridgroup/...

#golang #llama #ml

GitHub - hybridgroup/yzma: yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo.

yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo. - hybridgroup/yzma

github.com

October 23, 2025 at 3:08 PM

Reposted by The Hybrid Group

deadprogram

@deadprogram.com

"That Machine Always Lies: Truth and Fiction in the Age of Artificial Intelligence"

thatmachinealwayslies.com

That Machine Always Lies

Truth and Fiction in the Age of Artificial Intelligence

thatmachinealwayslies.com

October 21, 2025 at 9:59 AM

Reposted by The Hybrid Group

William (Bill) Kennedy

@goinggo.net

I'm so excited about the YZMA project from @deadprogram.com. I've taken 3 of his examples and cleaned them up. Next step is to build a mini version of the Ollama service to show the real power of YZMA. #golang

github.com/ardanlabs/ai...

ai-training/cmd/examples/example13 at main · ardanlabs/ai-training

Provide examples for Go developers to use AI in their products - ardanlabs/ai-training

github.com

October 20, 2025 at 6:23 PM

The Hybrid Group

@hybridgroup.com

Did we figure out how to use yzma to work on itself? We can't tell you that, but here is release 0.4 now with over 70% of llama.cpp functionality complete!

github.com/hybridgroup/...

#golang #ml #llamacpp #llama #vlm #llm #slm #tlm

GitHub - hybridgroup/yzma: yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo.

yzma lets you use Go for local inference+embedding with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo. - hybridgroup/yzma

github.com

October 14, 2025 at 12:43 PM

The Hybrid Group

@hybridgroup.com

We just released yzma 0.3 with embedding support, bug fixes, and more.

Go get it right now!

github.com/hybridgroup/...

#golang #llama #llamacpp #ml #vlm #tlm #slm #llm

Release 0.3.0 · hybridgroup/yzma

What's Changed feature: return wrapped errors when loading functions by @deadprogram in #17 docs: add information on what llama.cpp features are already implemented by @deadprogram in #18 examples...

github.com

October 14, 2025 at 7:01 AM

Reposted by The Hybrid Group

Takuto Nagami

@logica.bsky.social

#tinygo_conf
Thank you so much, @deadprogram.com !!!
(I added the caption, btw!)

October 12, 2025 at 2:01 AM

Reposted by The Hybrid Group

deadprogram

@deadprogram.com

Already have an update for yzma with Windows support working & also simplified developer interface for loading. Go get it now!

github.com/hybridgroup/...

GitHub - hybridgroup/yzma: yzma lets you use Go to perform local inference with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo.

yzma lets you use Go to perform local inference with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo. - hybridgroup/yzma

github.com

October 9, 2025 at 12:56 PM

Reposted by The Hybrid Group

deadprogram

@deadprogram.com

yzma is a new Go package for local inference with Vision Language Models (VLMs) & Large Language Models (LLMs) using llama.cpp without CGo.

github.com/hybridgroup/...

#golang #llamacpp #llm #vlm #slm #tlm

GitHub - hybridgroup/yzma: yzma lets you use Go to perform local inference with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo.

yzma lets you use Go to perform local inference with Vision Language Models (VLMs) and Large Language Models (LLMs) using llama.cpp without CGo. - hybridgroup/yzma

github.com

October 8, 2025 at 9:52 AM