porkychan

Mostly from runpod but going to buy some serious hardware soon cause costs are getting out of control.

I made a tiny local LLM 1.5B param (base model qwen 1.5B) and it beats Gemini 3.8 flash so a local 31B param beating fable is not out of the question. If you want to try my model and look at the benchmarks you can find them here: https://shlexball.com/