Look I have no idea what this is, but I am just playing around with some stuff that is being used to build the AI systems you are currently using. I am more interested in being able to replicate their technology using smaller (read: less funded) open source technologies that can run on my machine(s).
Managed to create locally generated embeddings. For the uninitiated, vector embeddings a way for machines to understand why Bill is William, and Richard is Dick. That's the easiest shortest description, but if you are South African, a better example would be: when I say the word Taxi, you understand we are not talking about a Taxi in how the rest of the world knows the word for it. For everybody else, embeddings, allows machines to understand language (at least words and phrases) in context.
I will update this randomly, I do not promise anything, but ya.
Lets assume I take this serious, (just remember it is 4:17am in Johannesburg as I type this), here is what I plan to tinker with, everything has to run locally,
- generate text*, image (audio?, video?) embeddings,
- train a min language model.
I am not a fan of python (this is the most python i've written in my entire life), so expect the programming language to change (rust, go, python, php, javascript, or) to whatever does what i need [to learn] at the time.
Yep, thats it. Feel free to do whatever it is you want to do with this.
1ove, creates.