I ran Gemma 3 4B without PyTorch, without CUDA — just flat binary files + 600 lines of Rust.
Paris #1. Jupiter #1. Ulm #1. Exact HF match.
CPU or Vulkan GPU. Any vendor. No framework.
Built on @chrishayuk's LarQL decomposition.
github.com/cronos3k/vindex-i…
#DeepLearning #GPU #MLOps #AIEngineering #RustLang #HuggingFace #Gemma #WebGPU
#CrossPlatform #EdgeAI #OpenAI #LLMInference #FOSS #IndieResearch
Replying to @chrishayuk
@chrishayuk watched your LARQL video this morning, spent the day porting it to Windows/Linux + CUDA and wrapping it in a Gradio UI — live on HF Spaces 12 hours after your release.
👉 huggingface.co/spaces/cronos…
👉 github.com/cronos3k/larql
Already running in 3 of my projects. Great work.
was messing with the OpenAI base URL in Cursor and caught this
accounts/anysphere/models/kimi-k2p5-rl-0317-s515-fast
so composer 2 is just Kimi K2.5 with RL
at least rename the model ID
so i figured out how gpt-oss does math... youtu.be/hbicn22nke4
just dropped my mcp client oauth library, which i think makes it easier to work with oauth mcp servers.. totally standalone not tied to any agent
framework.. github.com/chrishayuk/chuk-m…
so i updated mcp-cli... i think this version is a lot better, still got a bit to go, but i think it's much better
Super happy about this technique of using templates to bootstrap projects while vibe coding with Claude Code
youtu.be/V7YSfaQZrIw
updated mcp-cli to use gpt-oss as default model, works pretty sweetly.. also now supports gpt-5 and claude 4.1
recently spent some time with danny talking about pitching. was fun but different. youtube.com/watch?v=il4V23X0…
i've now moved mcp-cli onto chuk-llm with streaming, and really think it improves the performance now. i feel i'm getting close to a v1 release
i had some fun at the weekend, with my latest vid..
youtu.be/I-K3dVgKzGY
i made some big improvements to the a2a server over the weekend, it's now fully integrated with my chuk session manager. github.com/chrishayuk/a2a-se…. .
to complement the a2a cli, i've create an a2a server which makes it easy to host agents that are compatible with a2a. github.com/chrishayuk/a2a-se…
so i've created my first version of the google a2a cli... you can run it just using uvx a2a-cli
github is here: github.com/chrishayuk/a2a-cl…
Half way through the getaway, and it’s really bugging me to know which costumes @thewheatgerm designed, it must have been a few to get a credit….







