Hi there. I’m Josh, this site is where I’m keeping notes for my journey to understand LLMs. There’s a lot of nonsense I read about LLMs can and can’t do (mostly what they can do). I’m pretty skeptical of a lot of what I read. I can’t claim to be a proper skeptic if I don’t have proper understanding of this technology. I’ve grown tired of hearing the cheerleaders exclaim an LLM can do anything and everything.
The Blog is probably what you’re here for. If you want to start at the first page it’s here.
I explain a bit more of the why I’m doing all this in the About page.
The Setup
I have a Framework Desktop with 128G of RAM, which is running Fedora, which is running Ollama.
Tools
As I work on things, you’ll find them in the ai-skeptic repo.
TODO
This is a list of things I want to figure out. As I learn more about what I don’t know, I’ll add things here
- Model fine tuning
- How agents run tools
- Setting better context windows per model