README: add training corpus details

README: add SGLang config + real performance numbers

README: focus on the model, remove API endpoint

Nesi1 — uncensored AI for cybersecurity (initial release)

initial commit