Generative AI is undoubtedly incredible: everyday it is demonstrating phenomenal abilities and a proven potential across thousands of different scenarios. However, it also demonstrates the requirement of massive amount in computational resources to run: really impractical for many average users who would like to practice without investing thousands of dollars to just experiment with it and find the right tool.
My mission is to help driving the evangelism and the development of solutions that allow to execute simple, complex, and, why not, super advanced tasks on your hardware, without the need of cloud providers or bled yourself dry to buy a new GPU.
Embark with me on this journey to discover new tools, new algorithms, and new tricks to run Generative AI on prem. On your bare metal.
No expensive subscriptions or uncontrollable billing on your credit card.
I will focus on a few key concepts:
open source All the solutions proposed are freely available to you: to download, to try, to uncover, to replicate, and to improve with your contributions. AI should always be part of human knowledge as literature and science.
curiosity I want you all to be curious about new frameworks, new technologies, and new tools, even if they are in a very early stage and not ready for production. Wake up your mad scientist spirit!
local Local Training. Local Tuning. Local Inference. Local RAG. Going local means taking full advantage of your hardware, saving money and going green.
personal Personal also means "Out of the grid", so you'll able to also download the tools you need, allowing you to directly experiment in your homelab, with consumer-grade hardware.
private Private is synonym of secure. Your documents, your info, your secrets remain yours, residing in your local storage.
ethical Develop a responsible AI, balancing censorship and reducing the bias of the information without the influences of governments and big tech corporations. Make it yours.
community Fostering collaboration and international cooperation in AI democratization.
