Hello, I'm reaching out to share a project I've been working on and to ask for a hand with testing it.
I recently "vibe coded" a single artificial neuron. While it can serve as a foundation for neuroscience research, my primary goal is to build it up into a full neural network that could eventually run an LLM. It's very much an early-stage, exploratory project, but it's the seed of where I want to go. The project lives here: https://github.com/coralieayabie/hurd-translator-neuronne The catch: the project hasn't been tested yet. I'd really appreciate help setting up a testing environment with Debian, GNU/Hurd and QEMU — I'm still learning my way around that toolchain, so any guidance (setup steps, known pitfalls, config tips) would be invaluable. Going forward, I plan to keep evolving this project, scaling from a single neuron to a full network. To do that well, I'd love to learn more about: How neural networks actually work (architecture, training, backpropagation, etc.); How LLMs are built and run on top of such networks (transformers, tokenization, inference, etc.); Any resources, papers, or courses you'd recommend for someone at my level. All ideas and suggestions are more than welcome — whether about the code, the testing setup, or the bigger roadmap. Thanks a lot for your time, and looking forward to your thoughts! Best regards, Claire Ivanenka
