Abstract:
This talk shows you how to build your own local hacking assistant using uncensored large language models (LLMs, e.g. ChatGPT).

We'll build custom Retrieval Augmented Generation system (RAG) from 100 local PDFs about hacking, whip up an interactive app running locally and you'll walk away with skills.

Mainstream LLMs are useful, but they're also neutered. As a hacker it would be nice to get some AI hacking help without restrictions.
That's where uncensored LLMs come in.

We'll take a hands-on approach, showing you:
Local LLMs: Running them yourself (and why you'd want to).
RAG: Giving your LLM a brain (and a library).
Chainlit: Making it all interactive.
Demos:
OpenAI with RAG (the baseline).
Local LLM with RAG (taking control).
Uncensored LLM on RunPod (the real deal.
Building Your Own RAG: From PDF to interactive app in minutes.


Outline:
Intro: Terms, current state of LLMs for hacking
The Toolkit: Local LLMs, RAG, Chainlit
Demos: Building Unrestricted Assistants
DIY RAG: From PDF to App

Wrap-up and Q&A